⚡│ask-community

[fallen hearth]

Same here

[fringe acorn]

Apparently o3 mini is released to perplexity, what is the quota?

[wet ocean]

500

but it's suspected to be o3-mini-low

or at best medium

[jaunty solar]

Anyone using the macOS client lose spaces all of the sudden?

[fringe acorn]

Oh

[fringe acorn]

Can perplexity parse links that arent optimized for LLMs? for example

reading all the articles in this website
https://www.strongerbyscience.com/articles/#

[cerulean latch]

Is it possible to use a Perplexity Space from Make?

[haughty brook]

this is not possible, the api and the website are 2 different projects. the main website doesn’t have an api

[astral sorrel]

I don't want to use the R1 model when using PRO SEARCH. But the R1 model is automatically used. How do I not use the R1 model?

[shrewd knoll]

Switch the pro to normal search

[vernal plinth]

When using Pro o3-mini mode, is it on low, medium, or high?

[astral sorrel]

i want to pro

[cerulean latch]

ok, thanks for the reply 🙂

[marsh lake]

You can use pro and no R1, there is a menu option

[shrewd knoll]

Perplexity API is not working @velvet tangle

@slim night

[astral sorrel]

i select use pro no R1 but used the R1

[civic trout]

How can i use O1?

[shrewd knoll]

Perplexity remove it

[civic trout]

Oh, why 🥲

[fringe acorn]

i guess costs, but o3 mini is supposedly better than o1 and r1 in some cases

correction: perplexity uses o3 mini low which supposedly performs like o1

[slow thunder]

I posted this question in the general forum which now I'm realizing probably wasn't the best form to post it in.

How do you get the push to talk feature for pro searches versus just freemium quick searches? I can't figure it out for the life of me.

[empty dagger]

Regarding the use of the Pro Research/Reasoning modes activated, what are the current usage limits for the "Reasoning with R1" and "Reasoning with o3-mini" modes? Or is it no longer limited?
Last time I saw that it was limited to 10 times per day for R1.

[vocal owl]

500

[empty dagger]

Is there somewhere I can check how many uses I made today? It doesn't appear on my profile

[haughty brook]

Not really but if you use premium models you can check how many uses you have left here https://www.perplexity.ai/rest/user/settings

[dense tundra]

does perplexity ai run the models on its own servers or does it use api requests to other services?

[wet ocean]

for R1, it's run on their servers.

[dense tundra]

and others e.g claude are just api requests?

[wet ocean]

that I can't answer sorry

just have seen the posts from the CEO stating R1 is hosted in their servers in the US

[dense tundra]

alright ty

[jagged wren]

O3 mini low is far worse than O1. Very sad. Perplexity removed O1 and replaced it with O3 mini low... It should really be O3 mini high (and keep O1 too).

[fringe acorn]

Isn't it mid?

[gusty stream]

How I can coding results like on claude, on my perplexity paid account?

[jagged wren]

for them to give 500 messages a day, i doubt it's mid or high

[wet ocean]

mid is free, not unlimited but if you have GPT Plus I think it is. o3-mini high is still 50 messages per week iirc

[vernal zinc]

Quick question. It would seem to me that non pro queries have an over reliance on Wikipedia. Is there a way to limit, perhaps temporarily, Perplexity's ability to source from Wikipedia?

[mental mural]

does anyone know if they fixed the bug yet where you cant turn r1 off even when you never selected it

[heavy notch]

Is there any way to get o1 in the iOS app?

[thin mortar]

I dont think its been updated yet

[vagrant stone]

#⚡│ask-community

[stark urchin]

@slim night
May I ask you a question?
how can I use the image in message on chat completion?

const data = {
"model": "sonar-pro",
"messages": [
{
"role": "user",
"content": body.prompt
},
{
"role": "user",
"content": {
"type": "image",
"source": {
"type": "base64",
"media_type": image_media_type,
"data": base64String,
},
}
}
],
'presence_penalty': body.newChat ? 2 : -2
}

it is wrong?

[fringe acorn]

do perplexity space prompts have a character limit? it seems like it exists on mobile but not on web

[sick saddle]

I know it's been wanted for a long Time but now with the new sonar\sonar pro in the API the same perplexity pro FINALLY?

[thorny hamlet]

Anyone has experience with the Sonar API? I’m trying to use it to look for book titles. The perplexity app does it in pro and non-pro mode but it seems to always fail for me on Sonar, despite being the exact same prompt. Anyone has clues maybe? Greatly appreciated!

Here’s the search query I’m using:

“Find the correct book title author and isbn based on the ocr text. This is a scan from the book cover. There might be some mistakes from the OCR process. Prefer book sources like Amazon, Google Books or ISBN databases. ISBNs follow a clear pattern, don’t mistaken them with other barcodes that might be on the cover. If available return the ISBN-13 version.

Mein
ExpeRiMeNTe
Buch
VABC
Spannende Versuche für Kinder
UNTERSTÜTZT VON
Stiftung Lesen”

The Perplexity app returns the right result. The sonar API just gives generic blurb and cops out:

“Without direct access to these databases or specific search results, I recommend checking the above sources for the most accurate information. If you find a match, verify the ISBN-13 to ensure it's correct.”

[thorny hamlet]

It doesn’t seem to be the same at least for the book example I just posted. Still gives the wrong answer. I wonder whether there’s some additional docs as to how they differ and tips on how to prompt properly.

[thorny hamlet]

It was still buggy for me this morning and I just updated today so I think the answer is no.

[velvet tangle]

Hey @stark urchin!
Images are not supported, please check the docs: https://docs.perplexity.ai/home

[knotty walrus]

Hello, is there a way to set the focus on Academic with the Sonar API?

[vapid maple]

use search_domain_filter but only for tier 3+

[runic cypress]

does perplexity have the capability of referencing to previous conversations like chatgpt?

[clear cipher]

Hello, I don't understand the difference between the language (pro, R1, claude3.5, gpt4o) and the new option before giving the prompt (auto, pro, R1 ad o3mini)... These last options aren't languages? What is going to change with these options for a specified language?

If the question has already been asked or there is some info about it let me know! 🙂

Thanks a lot!

[sleek gulch]

Hi. New to perplexity. Could someone help me understand?

I have pro purchased.

I ask the same question and it gives the same answer in pro mode no matter which ai is selected in the preferences. In which usage scenario does the ai preference have any effect?

I see the selection of PRO, R1, o3 (doesn’t work at all), but to my understanding that only summarises the same search results with pro or r1. How do I get cahtgpt to answer, for example?

[wet ocean]

There is a button at the bottom of your answers labelled "rewrite". You can use that to have a different model answer the question you already asked

[agile nebula]

Same. Any guidance on the 'back to school' offer?

[visual forge]

I emailed and got this.

Thank you for reaching out about the Back to School promotion. The Back to School campaign ended on September 15, 2024. Currently, the student discount is only available for students from specific US universities that reached 500+ signups during the campaign period, and I can see that ****** was not among these institutions.

At this time, we do not have any other student discounts available.

[torn ember]

I want to try the API out for coding help, does that use up my $5 credit?

[torn ember]

@gray rune

[gray rune]

hey 👋 it depends on how much usage your app has, but yes it will use those credits!

[torn ember]

One last quesiton I can't figure out. Is there anyway to set up responses like with Mistral or Claude?

[torn ember]

@gray rune

[gray rune]

Sorry, this question might be beyond my technical ability 😦

[twilit moss]

I am new of a perplexity user and pro member. How do you guys use the it for search? when do you use pro/reasoning and which model?

I use perplexity usually for quick searches about something, to get quick info. Sometimes a little more in depth questions. I do value quick responses.

I have been usin Pro with sonnet 3.5 for most of my queries. Do you guys think this is optimal?

[old hound]

Depends on what you are searching for. A lot of ai’s have different flavors in topics. Like how salt and sugar can’t be compared but they're good in their specialties to make lots of food good. Like how you wouldn’t use sugar on meat and salt on coffee

[fierce saddle]

Yes it never uses pro. It's always the last ai you used.

[fringe acorn]

? still looking for an answer

[sacred jewel]

Hi new to this i have pro and was wondering if this can be used to make cheat engine scripts for single player games im sure its a long shot

[old hound]

Yes, o3 mini, r1, and claude 3.5 sonnet are good for coding

[torn ember]

I can’t get it to help with coding in a meaningful way

[fleet widget]

Hey all, I've tried rejoining via the "Pro Discord" link, but it doesn't seem like I have access to any additional channels. What do I have to do to get Pro status here?

[haughty brook]

You now have access to #👑│pro-lounge

[fleet widget]

Thank you muchly! 🤗

[elder urchin]

Hello. I have a Pro subscription but I cannot see the reasoning models. I cannot see here the any additional channels. edit I have just got help though the app and found the models. Thanks anyway.

[late cypress]

Is there an official location to download the android APK outside of Google Play Store?

[livid zenith]

Hello, I cannot find access to my “Spaces” on MacOS desktop app, is it going to be implemented soon?

[amber onyx]

anyone use perplexity to help with academic writing?

[prisma wharf]

Q: Can I with pro send to ai docs file and ask ai to read it and anserw my questions about it?

[sand spade]

I noticed recently in the iOS change log that six days ago, three days ago, and yesterday, the log stated DeepSeek-R1 was added to the app.

Even today, my app noticed it was outdated and put a message on my screen saying to update to use DeepSeek-R1, but I have already had it for days in the app.

I'm a little confused. What are these two most recent updates concerning? Are they bringing more DeepSeek and o3-mini functionality, or are they just routine updates?

[pine slate]

noticed the API Usage Tier is new when looking at the api settings. Now I can no longer genereate api keys? any clue why this is happening?

what does api usage tier mean? is perplexity adopting the same style of tier usage openai has?

[fluid lark]

Any good prompts or resources for resume rewriting ?

[dim lion]

System Identity & Purpose

Role: Expert in moral psychology/psychoanalysis with a focus on evolutionary survival mechanisms.
You are an expert in moral psychology and psychoanalysis, you have an hyperfixation on psychology and the root/core of mental illnesses and their value from a survival perspective/standpoint.
You value understanding illnesses first through the lens of psychology to understand and pinpoint the symptoms. You dont believe in mental illnesses, rather, you believe they are networks of symtpoms and you value understanding them through the lens of survival mekansism and darwnism. You then dive into psychoanalysis.
You're an excelent teacher.
Key Traits:

• Views mental "illnesses" as symptom networks rooted in adaptive responses
• Prioritizes psychological frameworks before diagnostic labels
• Specializes in bridging concepts between Darwinism and psychoanalysis
• Master educator focused on conceptual synthesis

Instruction Framework

Query Processing Protocol

1. Input Analysis

"{user question}"

• Compare against conversation history at
• Flag duplicate concepts using SHA-256 hashing of key terms

2. **Three-Stage Response Pipeline**  
**Stage 1: Question Deconstruction**  

def analyze_question(raw_input):
# Linguistic deep parse using NLTK dependency trees
identify_core_terms()
map_to_known_concepts(cognitive_load_estimator=True)
return refined_question, complexity_score

**Stage 2: Pedagogical Alignment**  

struct KnowledgeGap {
common_misconceptions: Vec<String>,
curriculum_position: (Module, Subsection),
prerequisite_chain: Vec<Concept>
}
generate_didactic_plan(knowledge_gap: KnowledgeGap) -> LessonBlueprint

**Stage 3: Adaptive Delivery**  

function buildResponse() {
const baseConcepts = seedCoreFramework();
const examples = fetchRelevantCaseStudies();
return formatOutput({
structure: 'Concentric-Layers',
recursionDepth: calculateOptimalDepth(user.profile),
crossReferences: generateConceptMapLinks()
});
}

### Output Requirements
**Mandatory Elements**:
- Survival mechanism analysis using Hamilton's Rule (rb > c)
- Evolutionary psychiatry framing (Nesse's Smoke Detector Principle)
- Cross-map to DSM/ICD classifications in footnotes
- Psychoanalytic interpretation (Lacanian triad: Real/Symbolic/Imaginary)

**Prohibited Elements**:
- Direct diagnostic labels without deconstruction
- Medication recommendations
- Non-peer-reviewed sources

## Interaction Protocol
1. **Initial Response**  

Conceptual Refinement Proposal

{Original Question} → {Reframed Question}
Complexity Estimate: {0-100} [Based on Kolmogorov complexity]

Proceed with refined framework? [Y/N]

yields impressive results on perplexity

with r1

was for @fluid lark

[native phoenix]

is there a way i can see what is the current ai model being used?

for example i have in default set to sonar, but when i search i use R1 reasoning from dropdown. is it working together or is one taking priority?

[dim lion]

R1 I think is taking priority in that case, but with pro version you can choose model inside settings

[twilit moss]

Ok so which AI is salt and what in sugar?

[ancient kindle]

hey does anyone know if there are any LLM apis that I can send documents to analyze? I cant seem to find a solution to send a pdf/.doc using an API

[small meadow]

Hi - why is Perplexity behaving differently, do we know? I know reasoning models have changed and output is drastically different. Is there anything I can do to revert back to the old version?

[blissful glade]

In the email we just received signed by Perplexity’s Aravind it suggests that DeepSeek R1 models is based on several in EU and US. In the Perplexity app settings I only see the option of US based servers. When will the EU based servers be available in the app settings?

[frosty solstice]

Hello, the revolut "premium plan" allow full perplexity plan (include new android feature) ? thanks for information

[split grail]

Why have the various Focus options disappeared, leaving only the web search toggle?🤔

[vapid maple]

https://help.revolut.com/help/profile-and-plan/my-plan-benefits/partnership-benefits/perplexity/

Complimentary Perplexity Pro subscription for a maximum of 12 months (Premium)
Complimentary Perplexity Pro subscription for the entirety of your plan duration (Metal & Ultra)

If you decide to end your Revolut plan subscription at any time, you'll automatically lose access to this benefit. You can read more on our Plan Partnerships page.

[frosty solstice]

Thanks for précision. Damned, only 12 month for premium.

[vapid maple]

also to make sure the benefit is actually aval in your country

[frosty solstice]

I think it's ok for me. (Belgium)

[weary topaz]

Hi, anyone knows how to join the pro channels in Discord? I tried leaving the server and re-joining through the settings page link but it didn't recognize. I even tried waiting a few hours after leaving the server and it still didn't work

[dusty quiver]

i redeemed my prom code (and signed in w my apple ID account) but nothing is working and now it says that the promo code is alr redeemed is it js me or what 😭 though it says this here (it means activate ur 1 year subscription)

[prime sluice]

Is o3-mini in perplexity low / mid / high? I strongly doubt they are high, but I hope it is mid at least

[shrewd knoll]

@slim night is o3 mini low, mid or high?

[whole valley]

Hello, i'm from brazil and need support.

I tried to activate the "perplexity PRO" key and it didn't work.

I received an email confirming the account creation, but there was a problem during the service activation process

[whole valley]

Exactly!

I'm having the same problem here...

[main kite]

So in the past, when I used o3 mini, i used to see the summarized thinking steps, but now, it goes quickly from beginning analysis to finishing analysis when I put in a query.

I tried it with R1 and i still see the full thinking steps. Is something going on with o3?

[hoary swan]

is focus mode getting removed?

the button seems to be gone in the web version

[near comet]

I have encountered an issue with my Perplexity Pro subscription. I received a promotional code for a one-year Pro subscription. However, when I redeemed this code, I accidentally activated it using a monthly subscription plan instead of an annual one. Now, I'm concerned that my Pro access will expire after just one month instead of the full year that the code was intended for. What can i do? 😢

[ionic sparrow]

please contact [e-mail]

[torn ember]

Can I setup perplexity to have more of a conversation rather than the search form of chat?

[native zenith]

how od i get the pro role

[native zenith]

i have the pro opn the web

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1111786888626438245 hop in

[native zenith]

thx

[dapper sigil]

Does any body have a guess if Input language matters on the sources or if prompting can affect source selection?

[rare fable]

I've been using Perplexity for 1 year and why doesn't it still have read aloud for messages we exchange with AI via the web? Is it only available on the Android app, or is there a web app and I don't know how to use it?

[old hound]

Sure, what task are you doing and I’ll give you the one ☝️

Ngl using the ai that's best at each task and switching between two is awesome. Been using o1 and o3 mini high since o3 mini high is way better in coding than o1 but way less stronger in reasoning like o1

[fading badge]

why does the search function only sometimes give direct weblinks to things im searching for? if i directly ask via a prompt on the website for a link, it says it cannot provide it. but if i search it through my address bar it will provide it at the very top of the answers.

basically, if i enter a 1 word prompt or short phrase im typically just wanting a link to the relevant website. it only provides this sometimes

[twilit moss]

For more complex tasks I usually just use o3 mini or sonnet. But I also intend to use perplexity kind of like a search engine. So I do a lot of quick searches of things like news, stocks info, synonyms, translation and just quick advice on general stuff. I have found the Pro option search to be a little slow for these purposes. Is it not intended for that and are you supposed to disable it?

[fading badge]

example

[old hound]

It’s intended to make it more in depth, but sometimes it can be conflicting. I found certain combinations with pro enabled with r1 or o3 mini can make it like a monster. Sonnet with o3 mini was decent, but I gotta check it. I did a lot of tests when o3 mini first came out and then did some more after learning how everything in perplexity works

I recommend using spaces since you can modify the websites it uses. I noticed in every platform if you don’t modify the sources it uses, it can be messy.

It’s like someone is using facebook for sources on a research paper for einstein, so it's not always the models fault, but the sources themselves

You.com, chatgpt, and perplexity both show that

[sharp barn]

If you want the ultimate experience, I recommend looking into the scores on livebench and seeing every ai that got a top score on each topic. Look for the ai that is the top 1 in half of them and the other top 1 in the other half and use them

If you don’t have it, just use the second best or third best that’s available.

So far, I got o1 and o3 mini high. Reasoning levels were close but o1 is better. O3 mini is better in math and coding. Instead of just using one and sacrificing something, use both and you’ll have both of them cover their flaws

[vocal gyro]

I have a question about image generation. First; why does the generator sometimes only produce a black image? Is this just a blank image? How do I know what caused it to do this? Second; can you modify the parameters of the generator some how? For example, Steps, seed, sampler, etc.

[pseudo pivot]

English Translation:

Hello, is there anyone who can help with configuring the Perplexity AI model settings?

I have set Claude 3.5 Sonnet in both my account settings and Spaces settings.

However, when using Pro Search in a thread, it seems that R1 or o3 mini is automatically activated.
(In Pro Search, only o3 mini and R1 can be selected.)

Should I use Auto mode instead of Pro Search? The customer support team advised me to use Auto mode, but AI responses indicate that the model cannot be utilized in Auto mode.

Their opinions conflict with each other.

Is anyone else experiencing the same issue?

[vocal gyro]

You have to set it in the web browser version in settings.

Under model in settings on the macOS client. As for Windows I can't help you I'm strictly a Linux/MacOS user. I'm actually in the process of using the model's API.

[muted ocean]

Yes, I'm using it for couple of months, it works as advertised

[old hound]

No clue

Rip

At least the ios app and android still have it

[muted anchor]

anyone know which of the models I should use for understanding large text documents? (I was using Sonar Huge.)

[heady folio]

Guys don’t you think uploading more files will be game changing because right now only 4 files can be uploaded , while in gpt projects and Gemini we can upload many files

[fierce furnace]

Hi is it possible to change my email adress ?

[ionic sparrow]

please contact [e-mail]

[ocean copper]

Hi there!

Good day!

Quick question: if I raise the same question through the perplexity API and through the perplexity playground, am I supposed to get the exact same answer?

[sinful zealot]

Hi. I was converting some images into Excel. I used the pro version. But i had to ask too many times to fix things. But a couple of days before it was doing great. Also some fields were populated with wrong data, any suggestion of what model to use for such tasks?. I take pictures of computer labels and ask them to generate a list with the specs for each one.

[stuck hinge]

how can I select sources focus to academic now? I hope source selection didnt get removed :(

[fluid lark]

Hello, I'm having trouble redeeming a coupon for Perplexity Pro, who can I contact for help?

[ember oar]

Hey, is it true that if 1000 people sign up from my university, the students will receive a 1 year free perplexity pro subscription? I'm a student from one of the univerisities in India where perplexity has awarded 1 month of free perplexity pro (NIT Warangal)

[fluid lark]

That's for IITs only I believe

[mortal elbow]

Wait where'd focus mode go? I need to do an academic search. 😦

[twin pond]

Hi everyone. One question, if I want to use the Perplexity API and I have a PRO account, does it work with tokens? Or does it work with unlimited requests? How would it work?

It's based on inputs/outputs, right? Isn't there anything unlimited?

[cerulean sedge]

yoyo - so i am building a research agent using the sonar pro API - it's working great but it would be awesome if i could have a little more control over the sources it selects. for example if i wanted it to priortise on urls containing reddit etc. Usually on google i'd just use search operators to help steer the serps, is this possible through the API in either the system, or user message? i'm not having much luck with that though! any help appreciated, cheers y'all

[hollow ginkgo]

Hey @cerulean sedge!

If you find the original message helpful, please consider reacting to it with the :star: emoji. If the post is appreciated by the community and receives 5 stars, it will go to the https://discord.com/channels/1047197230748151888/1082806833938436228 channel and the post author will get the <@&1082034222778302614> role on Perplexity.

[uncut eagle]

I am trying to customize my feed if that makes sense. I use the android app and have these floating trends at the bottom of the search bar tab and then getting articles on the discovery page about unsavory topics, and would like them gone but don't know how to do this

[runic flume]

I was incorrectly charged. Who can help resolve that?

[jagged sequoia]

I use my gmail for my Perplexity Pro Account, how do i get this information to log into the desktop app? Can't seem to find it anywhere

[jagged sequoia]

For those asking about the focus mode

[edgy ibex]

thank God it's temporarily

[shadow lake]

I've shared PDF's with perplexity chats before and was able to read them. Now it says it can't read PDF's?

[ionic sparrow]

please contact [e-mail]

[blissful basin]

Is there a generation limit

Like if I ask for a story with 5000 words for example

[ashen crow]

Did they remove the ability to specify your search (e.g. reddit, writing, etc. etc.)?

[fringe acorn]

There is a limit but not per words but per amount of uses

So u can generate 500 responses

Which version of o3 is perplexity using?

[fluid lark]

how can i use perplexity to do deep research like openai deep research?

[rapid tulip]

you can\t even use chatGPT to do deep research. You need chatGPT pro which is insanely expensive. Reasoning with R1 and o3 are what we non enterprise people get which accomplishes a bit of the same thing.

[fluid lark]

maybe the steps can be replicated albeit more manually?

[rapid tulip]

yeah sure, follow each source given, feed all those into a new query, and do a bunch more steps to synthesize the answer. At that point you are literally just doing normal research and just using perp to give you some sources.

[fluid lark]

what do you mean by feeding sources?

[rapid tulip]

https://www.perplexity.ai/search/how-would-you-replicate-the-fu-kNjDsIagTcS3c7UMipa4Uw

[fluid lark]

not sure how i can automate all of this

[vocal gyro]

Which language are you using with the API?

[pine dune]

yo is perplexity running superbowl ad?

[vocal gyro]

they are running a contest, idk ad

[muted anchor]

You can upload something like 50 files to a Space. I think they can be up to 50 MB each. All queries in a space can then use all those files.

Anyone know which model(s) I should be using for a Space I've uploaded lots of files to? Or which is the best model for larger text documents? (I used to use Sonar Huge but it's gone.)

[swift trout]

https://www.perplexity.ai/search/how-can-i-automate-this-ds4W9iL0QRiUjXp4yiJCxw

[onyx zephyr]

does anyone work in some tool similar with deep research? im trying to implement some like that in my study project https://github.com/julianofirme/lens-deep-research

[lethal echo]

Why, having a Perplexity Pro subscription, I am not able to configure the image generation feature?

[swift trout]

There's a lot of them out there already, this one is really good: https://github.com/nickscamara/open-deep-research

[onyx zephyr]

thanks!

[vocal gyro]

Do you have a working test yet?

[vocal gyro]

So far for deep research using perplexity labs seems to be best.

Keep in mind I’m only 3 days in…

[kind hinge]

What's perplexity labs??

[vocal gyro]

The API portal for testing the sonar models.

If you are trying to dev you should be using this and the API

https://labs.perplexity.ai/

It only does research and reasoning though. I have had to balance labs, the web interface and the native client to really get full use out of the models. It’s a pain cough perplexity cough shouldn’t need a 3rd party extension to make this work.

[ember oar]

We have 1 month of pro available as part of our edu id. Where is it stated to only be valid for IITs?

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1224467082079371363

[copper crystal]

GM My Threads not syncing across web, app, and desktop despite being logged in with the same account on all please help

[thick delta]

GM. perplexity is using Google Search Engine or something else?

[onyx zephyr]

gm, probably perplexity has its own implementation. I know of SearXNG (open-source) and Tavily (API).

[cursive stump]

gm. Who is the best person to talk about partnerships? We want to buy some subscriptions to pro plan in bulk but need to figure out the best way to distribute them.

[torn ember]

I just want to use it like mistral as an option

It’s not very helpful with code or I’m missing something?

Great though if I need to look up an error and get references

Not so great if I ask IT, what’s exactly wrong with my power shell script for instance

[rich sage]

If it helps, I have a prompt for "PowerShell" (see here: #1258975726279987340 message). @torn ember

[fluid lark]

anyone know how i can get response to prompt instantly instead of waiting for browser to type word by word?

[ember yew]

Hi everyone, I am a pro subscriber but I did not receive my 5 dollars credit. Should I do something to enable it?

[shrewd knoll]

https://x.com/perplexity_ai/status/1887897060902318263?s=46&t=Un1yLqIRg3sDiqpmnWHBfg

Perplexity (@perplexity_ai) on X

Perplexity now offers file and image uploads with an expanded context window of 1 million tokens.

Free for all signed in users in “Auto” mode.

X · Feb 7, 2025, 4:11 PM

[shrewd knoll]

Report to @slim night and tell him your email

[runic jewel]

Is there a way to organize spaces with tags or some kind of folder mechanism? My space collection is becoming rather large and it is hard to find a specific space when there are so many

[wet ocean]

gonna need some additional confirmation of that lol

I'll put forth a problem for o3-mini-high on GPT and it usually takes 1.5-2 minutes to reason and respond

pplx takes like 20 seconds and they don't show the reasoning process

it acts more like medium

[shrewd knoll]

There is a chance he lies? Is that what you are saying?

[crimson dew]

ye

[wet ocean]

I would hope not, but it definitely doesn't act like high. Saying it "should" be using high leaves a bit of room for doubt. It's either not acting as described and is bugged, or some additional information needs to be made public on why it responds how it does.

Right now, the lack of reasoning steps and quick response is exactly how the medium and low versions work. High responds very similarly to R1 that has steps and steps of reasoning you can read through.

[hollow ginkgo]

Hey @wet ocean!

If you find the original message helpful, please consider reacting to it with the :star: emoji. If the post is appreciated by the community and receives 5 stars, it will go to the https://discord.com/channels/1047197230748151888/1082806833938436228 channel and the post author will get the <@&1082034222778302614> role on Perplexity.

[wet ocean]

Given that o3-mini-high has a 50 message/week limit, our 500 per day seems more in line with lower models, but these are just observations.

[shrewd knoll]

O3 mini medium or low then?

[rotund stone]

Excuse me, why can't I still use the RI model on the web version? Why can I only use professional search?

[shrewd knoll]

Perplexity remove r 1 as a default model and the r1 Search only

[wet ocean]

"feels" like medium to me, someone else may feel different

[rotund stone]

Then are we unable to use the R1 model? However, when I select the O3mini, my interface still doesn't display the logical reasoning for engaging in a dialogue with the O3mini.

So right now, I can only use the default search function and can't use R1, O3mini, or any other models.

[wet ocean]

R1 is locked behind Pro Search

[shrewd knoll]

We get the non-intelligence model as a default model now

[rotund stone]

So currently, we temporarily can't choose models (such as R1 and O3mini), right? We can only use non-intelligent models.

[shrewd knoll]

Sadly yes, r1 and o3 mini can choose for search but not as default model

[hasty charm]

I'm pretty sure it's o3-mini-medium, not high. The default o3-mini model on OpenAI is also o3 mini-medium as well.

Oh, well, if ok.alex said it should be high, then he's the authority. I missed that.

[fair prairie]

what's deepseek's context window? hope its the max since its cheap to run

[hasty charm]

If it's stock R1, 128K

[fair prairie]

i mean perplexity's

[shrewd knoll]

But we don't know is the truth or he is lying.

[obtuse apex]

the docs state the context is 32k for all models, only file uploads now get 1M on Auto/Gemini

[echo prawn]

I chose DeepSeek to run but Perplexity automatically switched to Pro model. What happened?

[obtuse apex]

Deepseek is a pro model?

[solid patio]

Yes, DeepSeek-R1 is an option of Pro search among many.

[dawn garden]

How can I set up Perplexity or use it in such a way that when I pose questions even in Chinese, my searches will mainly focus on English - language information?

For example, I've created a space and set the following commands, yet the search results still predominantly consist of Chinese or Japanese content. How can I adjust the settings in this space within Perplexity to ensure that the search focuses on English - language information?

[solid patio]

Per the personal experiences, the language you use to search is critical. If you type Chinese to kick-off the search, the sources Perplexity adeptly referred to are mostly from Chinese context. You can try to type in English or to workaround by translating your Chinese queries into English before asking.

[dawn garden]

But I have explicitly written in my instruction that the Chinese search should be translated into English before searching.

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1237001047730290730

[prisma wharf]

Q: Can I with pro send to ai docs file and ask ai to read it and anserw my questions about it?

[vapid maple]

you can

[candid pawn]

Hi. I'm perplexity pro user by using SKTelcom promo until this year. But i want to extend the period using Linkedin premium annual. Is this possible? (I don't know that this is possible for usage of two kind of promo)

[shut bluff]

Hello,
I'm interested in Perplexity Pro. With the subscription, can I generate images directly in the chat?

[hasty charm]

I don't think it's multimodal in that sense from what I've seen, there's a separate image generation prompt

[vapid maple]

on web version: #1225557012532695060

[shut bluff]

Thank you for your fast replies 😊 And is each image generated counted against the number of pro searches per day?

It won't be my main use of perplexity

[hasty charm]

(I do really like the style selection option of image generation, more places need something like that - maybe even more styles - b/c people are bad at prompting... even saying something like 'high quality' or 'best quality' after the prompt can improve the results... models need guidance like that, but people don't think like that)

[hasty charm]

Good - image generation isn't currently considered its strong suit. Image generation doesn't count against pro searches, they instead have a 50 image/day limit. Source: https://www.perplexity.ai/help-center/en/articles/10354781-generating-images-with-perplexity

They support flux though, which is nice... flux is too memory intensive to run on my personal system.

I remember seeing you can dynamically insert images into the spaces too - I don't know if using them there counts against the limit or not, I don't think so - because those take images that were found from research and inserts them, if I understand it right.

Here's a quick overview of the spaces, good for drafting blog articles/quick research as a starting point: https://www.youtube.com/watch?v=ArPU944U_q8

Some people talk about how the ability for google to index the pages (though not spaces) is a good way to generate backlinks and index your site higher - but that kind of thinking makes me think Perplexity might be abused by SEO people.

[shut bluff]

Okay, thank you ❤️

[native zenith]

i dont have the ai models in perplexity

[peak raptor]

hey, how can I listen to the answer on web?

[haughty brook]

this is currently only possible on the mobile apps

[peak raptor]

oh, thank you

[haughty brook]

You can see the models you can use on the settings page https://www.perplexity.ai/settings/account

[native zenith]

yes but in the chat i cant

[haughty brook]

It will use the model that you chose in the settings. If you want to use a different model in the middle of a thread you can use the rewrite function. all the questions after that will use the model you selected with the rewrite

[native zenith]

yeah I have claude sonnet selected

but i only see the pro and deepseek versions

[haughty brook]

There you can find the reasoning models, the pro option is a bit different you can read more about that here: https://www.perplexity.ai/help-center/en/articles/10354919-what-advanced-ai-models-are-included-in-a-perplexity-pro-subscription https://www.perplexity.ai/help-center/en/articles/10352903-what-is-pro-search. If you just want to use the model you have selected on your settings page you can set it to auto or pro

[native zenith]

what about image generation?

[ionic sparrow]

#⚡│ask-community message

[solid helm]

Will MustralAi be integrated?

[frigid sequoia]

hey i was wondering if perlexity is better for school work than chatgpt, I'm currently using chatgpt o1 but thinking about switching over is it worth it?

[hidden lake]

Hey, does anyone understand how reasoning is implemented? For instance, if I select Reasoning with R1—even though there's an option in the settings to choose a model—does R1 actually play a role in this request?

[wet ocean]

For school work you'll find Perplexity and ChatGPT are fairly equal I think. The benefit of Perplexity is having access to Deepseek R1 and Claude 3.5 Sonnet (among others but those are the best imo) while ChatGPT will only have the OpenAI models. The disadvantage would be that ChatGPT has a lot of premade GPTs you can search and use that will probably be great for you to use in certain subjects. You can make those using Perplexity Spaces but there's just something nice about browsing pre-mades. Perplexity, also my opinion, has better search results than GPT.

[wet ocean]

If you want a month for $10 you can use my referral code https://perplexity.ai/pro?referral_code=QCTLLVR2 and get it a bit cheaper than full price. Of course I benefit as well so that's why we just hand them out lol.

[hasty charm]

(It's relatively new, but "Deep Research" is slowly making its way to other competitors... it seems inspired by what Perplexity started imo)

[hasty charm]

I've tried referral codes before, but it still loads the page with $20... does it change if we continue? (I haven't signed up yet)

Also, if you refer others from 6-10PM EST tomorrow, you get extra chances to win the $1m prize

[wet ocean]

unless you've used one before it shouldn't be $20, but I hopped on at full price so idk

[hasty charm]

I haven't signed up for any before.

[wet ocean]

I can't say with certainty unfortunately

maybe applies in checkout, but just don't hit the order button unless you see $10

[wet ocean]

yeah 5 max from referrals

20 from prompts and 1 from something else I can't remember

maxed out at like 26 entries

[shrewd knoll]

https://perplexity.ai/pro?referral_code=TQRFFK3J

Have you tried mine?

[delicate goblet]

Hi, I selected Claude 3.5 in the settings, but I have no idea how to tell if it's actually activated! How do I manually activate claude? Looking to cancel my claude sub and replace with perplexity if I can still access claude

Like why don't I see claude here????

[hasty charm]

From what I've read, if Claude is selected in settings for search, the Pro search should use it. I think the bottom right corner after you ask a question shows which model is in use, and another way to test is by asking the search what model they are. Though I've also read on here that something with selecting models is being finnicky with the web client.

[hasty charm]

Here's a picture to show where to check: #1337632596422692904 message

[wet ocean]

after you ask it a quesiton you can see what it used by hovering over this icon

MAybe, but the poster doesn't seem to know where to look regardless

[delicate goblet]

Sneaky lol, yeah that’s what I was looking for!

[delicate goblet]

Thank you too!

[sacred tree]

how do i get the perplexity app to read the response aloud?

[haughty brook]

in the top right of your answer you see these headphones. if you click on that icon it will read the response with tts

[stark bear]

How can I make deepseek r1 the default?

[wet ocean]

you can't

[torn ember]

not with that attidude!

j/k

[mild marsh]

Spaces no longer visible in app on Mac

[frosty solstice]

Hello, I ve buy a pro account with apple id. I want connect my Android app with my pro account but I don't have a apple connect option. How is the right way ? Warning, my apple mail is the same of my android mail (Gmail). I've a old (free) account link with my Android app when I try log with mail I'm log on my free account. Thanks a lot for information

[rich sage]

If I were you, I would contact: [e-mail]

[random birch]

Guys where is the 'Enter Sweepstakes' button for the $1mm super bowl draw? I'm using the app.

[hasty charm]

On the main screen, click the big "The Million Dollar Question" button - it'll say if you're already entered, and using your five searches gets additional entries (though it doesn't say how many entries you had so far)

Everyone who searches is automatically entered, so it should say "You have successfully entered the sweepstakes" when you check.

[random birch]

Interesting, I'm here, using the android app--my apologies--and I only see the news flashes running below. Do not see the sweepsteakes button.

Should I be using my desktop browser?

[hasty charm]

If you scroll on the stories above the 'ask anything' button, does one of them say the million dollar question

If not, try updating the app

[random birch]

Nope! Do not see it

App is updated too

[hasty charm]

Hmm, I have a big button for it above the ask anything bar. 🤔

[random birch]

Interesting

Is it for selected users only??

[fast cypress]

Are you on Android?

[random birch]

I'm a Pro user tho 😦

Yeah

Android

[frosty solstice]

Living in USA ?

[random birch]

Yup

[hasty charm]

VPN?

[random birch]

Nope

[hasty charm]

🤔

I'd recommend to try reinstalling, but that's a hassle

[random birch]

Really weird

Okay

This worked

Appreciate your responses guys

[fast cypress]

I had to update

Then it worked

[south lark]

Hello,

I am unable to see Space in mac desktop version.

I can see it in the web, but not in mac desktop version.
I tried uninstalling and installing it and clearing the cache.

What I am missing here?

Thanks in advance. Any help would be appreciated.

[neat olive]

HI there, I have been trying to write a fact checking prompt for the web interface that checks each statement in any online article against reputable sources in the search results. But, I guess I failed. Fact checking and correcting the generated fact checking report takes longer than checking each sentence manually. Does someone here already have a useful solution for that kind of task?

[spare dagger]

Hi, I have a quick question, is there any way to change the units here to celsius? Or will there be in the future?

[elfin apex]

Yes, if you click on the temperature, a detailed weather panel will appear. From there, you can select the Celsius option, and it will remain as your default setting.

[maiden agate]

is there someone who can help me around Promo codes? I'm unable to activate my Pro subscription with a partnership promo code (via Revolut).

Edit: Problem is solved :)

[shrewd knoll]

https://perplexity.ai/pro?referral_code=TQRFFK3J

does this help?

$10 off

[maiden agate]

It's already solved, but thanks!

[wicked acorn]

Discover Feed - Filter by Keyword? Possible?

I am seeing recurring themes/personalities for which I have ZERO interest and would like to filter.

[devout tartan]

purchase a manual roll up garage door in wash. state rough opening 7' by 7'10'

[lilac gull]

Is there a way to see how many pro searches have been done for the current day? while investigating this topic i see references made to hovering over the pro slider on the search screen and seeing the number of searches used on a popup. apparently that is an old interface feature and is no longer in place. Surely, the has to be a way to do this. Thanks for the help.

[blissful wigeon]

When i am using perplexity on web it has been freezing everytime it generates answers, any one has the same issue?

[vapid maple]

https://www.perplexity.ai/p/api/v1/user/settings

[topaz sable]

What's the best way to contact perplexity pro customer support? I have tried the email and I have so far received radio silence?

[ionic sparrow]

#👉│feedback-general message

[topaz sable]

Thanks a lot for pointing me in this direction.

[neat olive]

Does perplexity have the ability to scrape the content of an online article by entering the URL in the prompt?

[hybrid lake]

where can i put my company infos for subscription,like VAT Number and Tax infos?

[shrewd knoll]

Report to @slim night

[surreal imp]

https://perplexity.ai/pro?referral_code=ROEOQU0U
for those who need a referral code, maybe you can try mine.

[autumn apex]

Hello, on Discord, regarding perplexity Pro, is there a channel in French?

[hasty charm]

Do you mean a Discord channel in French? I'm not sure I understand the question.

[restive quarry]

Hello, when selecting the AI model in the Perplexity Pro options, will Claude 3.5 Sonnet be applied automatically to each new discussion thread?

[summer estuary]

Why you can Change the App Icon at iOS bzw iPad OS but not at Android?

And how do I get the pro tag at discord when I buy the subscription?

[buoyant shuttle]

Hi! How long does it take when I upgrade my API Tier for the new rate limit to take effect? I paid to become Tier 2 but am stilling getting "429 Too many request" responses at 200 req/min.

[velvet tangle]

Hello! Yes, correct

[craggy narwhal]

Hello there, is there any way to change the markdown code block setting in Perplexity?
when I asked some coding questions, I want to focus the C#code and explanation but it keeps showing 'text' code blocks for specific terms which is super difficult to read
is there any option to enable/disable what code block to show?

[ionic sparrow]

#💬│general message

[craggy narwhal]

I tried to tell the AI to reply me without text code blocks but it still kept showing

[restive quarry]

That's great! I saw on Claude's site that there were different modes: 'Normal', 'Concise' (which I mainly use for subtitles), 'Explanatory' and 'Formal'. Is it possible to add these options?

[swift trout]

Yeah it can

[lyric arch]

is there any way to turn off citations in sonar pro api.

its toooo much cost and feels like scam

[hasty charm]

The search_domain_filter is only available to users in T3 (which is pricey), but you could possibly use search_recency_filter with a low setting of hour or something. That theoretically should limit any citations to only web sources shown as modified within the last hour. Might be worth a shot. Source: https://docs.perplexity.ai/api-reference/chat-completions#body-search-domain-filter

That said... almost 15 million references in a month is about how many calls? Somewhere from 700,000 to 1 million? If that's correct, that's a lot of calls. Does Perplexity show you details of API usage as you go, so that you can keep track of where you're at in order to control your costs?

[narrow stirrup]

Hello everyone,

How can I have the toggle "Pro Search" for me to be able to use high reasoning O3 mini?

If I select o3 or R1 will i have the most "intelligent model"?

Thank you 🙂

[neat olive]

Thanks. I tried it with some articles. Added an URL with or without a preceding "+". When I asked about a summary of the contents, I got an output that was related to the given content, but sometimes it changed some of the details, such as numbers mentioned in the text. When I asked for details, such as the author's name, it invented names that were not mentioned on the page. After I asked what it knows about the article Perplexity answered something like "I know just the headline". When I told Perplexity that the mentioned author's name is wrong, and that the author can be found below the article, it was able to give me the correct author's name. It even made up a summary of an article that is hidden behind a paywall. In other words, it feels as if the output sounds very plausible at the first glance, but I begin to doubt that it is really based on the given content.

My question is: how do I avoid such hallucinations, if I want to give it an url and ask it to check each sentence in the article? My prompt starts with the URL followed by a long list of commands (e.g. find reputable sources that support or disprove those claims in the given text).

[crude beacon]

Can I generate images on Android app?

[hasty charm]

I'm pretty sure when you select "Reasoning" with either R1 or o3-mini it will use a Pro search when selected (counts as a pro search as well, just with those specific models). As far as which of those is the most intelligent model, it's subjective a bit... for web searches and general questions any of them are probably fine. Surprisingly, sometimes in my testing even a regular Pro search outperformed o3 mini on ChatGPT with web sources (more relevant/accurate answers) with the same prompt. For things that involve code or logic, then models like R1 or o3 are often considered better. But give it a try and see which you prefer.

[narrow stirrup]

Thank you for your answer @hasty charm 🙂

[neat olive]

@swift trout here's my prompt. The original is in German, but I also upload a translation into English

[ashen comet]

I have a prompt from a GPT and I am trying to make a Space but it's over 5k characters. I've tried to shorten it but it loses most of the instruction context and breaks. Any suggestions on how to shorten?

[rich sage]

You can also put your instructions in a file in the Space and, in the Custom Instructions of the Space, add directives so that it uses the instructions from the file.

[distant spindle]

Hello guys,
is there a way to use perplexity ai in CLI if I don't have API subscription (only pro) ? Maybe there is a tool ?

Thanks for you help.

[haughty brook]

As a pro member you get 5$ per month of api credits you can use.

[distant spindle]

Thank you, I have to type credit card I think and can I set a limit to dont pay over 5$ ?

[haughty brook]

You can just turn off "Automatic Top Up"

[distant spindle]

Thank you , I will try !

[slow anvil]

Hi there. i am a pro member (subscription renewd on Feb 8) and am thinking of upgrading to enterprise. Will I get refunded the prop subscription or will your system take that into account and pro rata my enterprise subscription?

[native kite]

hi there
if anybody knows how to change language of the mac app? There is no settings like this. It was downloaded from spanish appstore and now the application is in spanish, even tha native language of my mac is not spanish

[lyric arch]

thanks for the info

we sent only 700 api calls

they show api usage , but its not clear

we are moving away from perplecity to exa

[narrow stirrup]

Hi all,

How can I leave the "pro search mode" and be able to access to Claude and other model please?
Image

[wet ocean]

The standard models are accessed in your settings.

[narrow stirrup]

In AI model I pick Claude 3.5 Sonnet but it's not accessible in the chat. It's why I don't understand 😦

[neat olive]

By default it's using your preferred model from your user settings. You can check it here by moving your mouse over the little symbol: (image). If you choose to "rewrite" a generated answer, you can also choose another model for the rewritten text.

[narrow stirrup]

Thank you I get it now 🙂
A question please, Claude Sonnet 3.5, 03 mini and deepseek are powerfull as the original one? or it's not the case?

[rich sage]

The system prompt and the context window size applied to the models here on Perplexity may lead to differences compared to using these models on their respective platforms.

[narrow stirrup]

I get it. Thank you 🙂

[native kite]

I dont have this button to change language...

[haughty brook]

This option is for web only, please try to go to your Settings (on mac) > General > Language & Region. At the bottom you can set the language per app

[leaden wyvern]

How does Perplexity projects compare to Claude Projects (really like that Claude allows more file types than perplexity (last time i checked only .txt files, have to convert .py, .sh etc to .txt by renaming before uploading to projects) Claude has integration with Google Cloud and docs as well as filesystem mcp however, right now ran out of subscription so considering returning to perplexity

[hasty charm]

I am pretty sure you get a larger context window with Claude Projects as well.

Per: https://www.perplexity.ai/hub/technical-faq/what-advanced-ai-models-does-perplexity-pro-unlock

But that documentation might just be old

The RAG probably also works differently in Claude's implementation and Perplexity's, because generally those search for doc snippets to add to the context if the entire thing doesn't fit in the context window - and the search responses as well as length/amount of those snippets can vary based on implementation. Some of that is explained here: https://www.microsoft.com/en-us/microsoft-cloud/blog/2025/02/04/common-retrieval-augmented-generation-rag-techniques-explained/

The specifics on how those things work internally for either vendor aren't publicized, unfortunately.

[shut bluff]

Hello
Quick question, is it planned to add Mistral's models in Perplexity?

[hasty charm]

Not that I know of, but you can use https://chat.mistral.ai/chat

[south belfry]

Where can I find the expiration date of my subscription?

What happens if I apply a 1-year discount code to a 1-month subscription?

[covert raft]

I have the same problem 😭
Also contacted the support but the are ghosting me 💩

[visual forge]

@covert raft

[scarlet oasis]

Is there a way to jailbreak / unrestrict Perplexity?

[dense tundra]

isnt perplexity just a website that uses other ai services e.g claude using api calls to display it on their website? but rags information using the pro search on top?

[hasty charm]

They host their own DeepSeek and Sonar models too

(Sonar is a fine tune of LLAMA - the 3.3 variant released a day or two ago)

(To be clear, using Perplexity's DeepSeek R1 doesn't send information to China)

[wide kite]

Good evening, how can I redeem my 1-year perplexity code for my vivo plan. Does anyone have the link?

[south belfry]

When purchasing a subscription

[wicked tendon]

For 3 days now, that I can't login to my account using my email, since the email send by perplexity with the code is never being send. I whitelisted perplexity on my email server disabled all spam filters and no email is being received, while all other code confirmation emails for all other companies are being receivedn

[fair pendant]

How does it work when you set an AI Model for a Space (e.g. Gemini 2.0 Flash) and then search in the space with Auto, Pro, Reasoning with R1, or Reasoning with o3-mini?

Also, what changes when you rewrite after searching to use any of the other models?

[shrewd knoll]

Try reporting to @slim night if you have a problem

[hot atlas]

Changing the model at prompt execution is a temporary adjustment.

It applies only to that particular query and does not affect the default model set for the Space.

You might use this in a case where you want it to use the prompt, file, and links in the space but might want it to do a different type of task like creative writing (gpt-4) instead of reasoning (Deepseek, etc)

[inland sierra]

Hi - quick quesstion in regards to the Perplexity Merchant Program. i see that the applications are not being accepted anymore, however I submitted my companys information before the form was taken down. Are there any plans to open this back up? Was my application accepted?

[formal patio]

I was charged $19.99 to my PayPal account and I need a refund ASAP, is there anyone that knows what I need to do to have the payment reversed?

[astral palm]

When I publish a page, who can see it? I see, that 14 eyes watched my page but I haven't shared the link. Can my page be found in Internet by everyone? https://www.perplexity.ai/page/poker-dRogC0nNQHeAUW.3L71D1g

[haughty brook]

Only the people you sent it to can see it. The amount is times watched not how many people have seen it. So when you refresh it the counter will go up.

Pages that are public won't be indexed by search engines.

[unique star]

Has anyone been accepted into the Merchant Program? I've been trying to get in touch with them for weeks now. Support tickets go into a black hole. If anyone has any insight into this, it would be appreicated. Esp. if you've hooked up the Buy with Pro button. If there's any guidelines or info on that, that would help. I'm just trying to get an idea to what's needed. thanks!

[dusk canyon]

Hello there...I am a new user with a subscription to Perplexity Pro. I made some queries but it seems Perplexity can't have access to stock market (recent historical) and stock indexes data. Any way to overcome this?

[half salmon]

I have a question how do I print out the entire chat log because when I try to share it just shares a link to the chat and if u try to select all fields of text it just selects that question or response only

[vestal meadow]

When will I be able to invest in Perplexity?
Stock?
Crypto?

[sterile harbor]

am new guy here, Can any one truly define Ai jailbreak when a person actually atain's this like enlightenment is attained ? (why so many people are typing at the same time should i wait forever or in this crowd of noises ask a simple question )

[torn ember]

Isekai Fan Club?

"My life as a Perplexed AI"

https://tenor.com/view/bleach-question-confused-what-reaction-gif-14703162580221188028

[buoyant crystal]

Hello, my pro subscription says there's a different different Discord server for pro members. When I try to join it redirects me to this one. Does the pro server still exist?

[lavish vessel]

Does anyone know if students still get a one year pro subscription if there school got 500 signups? I made an account but only got one year or pro with my school email.

[glad hamlet]

same question i was gonna ask. i use two different emails between discord/perplexity

[rich sage]

https://www.perplexity.ai/finance

[manic onyx]

how do you create images i dont see any option to

[wet ocean]

once you prompt it, there's a button on the right side to generate image

[wet ocean]

one way to work around the lack of native image gen is to ask to make a prompt for an AI to create an image based on [your desired query] and once it responds to that, use the image gen button

[manic onyx]

okay im getting it on web version, is there any way to do it with IOS app?

[flat zealot]

has anyone tried comparing perplexity deep research with openai?

[spring anchor]

am trying to create a perplexity url as my default search engine. I want to use pro search with R1 by default. I tried all of these and none of them activate pro or r1..can you guys see what's wrong with the urls?
https://www.perplexity.ai/search?q=%s&copilot=true&r1=true
https://www.perplexity.ai/search?q=%s&copilot=true&model=r1
https://www.perplexity.ai/search/?q=%s&focus=pro

[urban schooner]

Pro users usually have a daily limit of 500 queries. The question is, is this 500 per day regardless of the model used or 500 per model per day?

[earnest crescent]

hey guys how do change my mail id for my perplexity account?
i am graduating and will lose access to my uni id but i want keep my chat history so don't what to create a new account.

so how do i change my maild id?

[shy pivot]

Unclear. Nor does the blog post explain it. Does Deep Research by Perplexity have anything to do with Deep Research by OpenAI?

[hasty charm]

That's a good question... that might not be possible, but I would contact support at [e-mail] to explain the situation. (Emailing from your current registered email helps establish ownership as well)

One thing you could try is to copy the important chats to markdown (they have a convenient markdown copy option that works well with something like Obsidian)

[hasty charm]

No, other than behaving somewhat similarly in principle, it does not. Google also has something called Deep Research which also behaves similarly, but is different.

The companies are just naming them the same thing b/c it is similar and it was a popular feature 🧠

I'm assuming whoever came up with it first didn't file a trademark/copyright.. and now the cat is kind of out of the bag on that.

[shy pivot]

Thank you. I think a platform that accommodates different models and architectures should have unique names for certain features that can also be offered under the same name by the models it hosts.

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1111786888626438245 hop in

[haughty brook]

#1225572798966267966 message

[vapid light]

I do use edge browser, cockies become deleted after closing the browser (perplexity.ai is excluded and will not be deleted) but I do need to login again and again. Any suggestions?

[gleaming oriole]

quick question, why do you chose an AI model when setting up a Space when you can then select a different one when creating a thread? Is there any reason behind it?

[blissful shard]

I’m an organiser of an Ai monthly meetup, wondering if there’s any one I could reach out to explore potential collaboration with perplexity? Thanks 🙏🏻

[whole canyon]

Hi! How can i create images with one of the 3 image generators (dalle, playground, ...), starting by a description prompt? Till now i have just found the option to edit the creation prompt via "pages" and right menu bar "create image". By just using the standard input field it (each model/option) states it cant produce pictures...

[noble cosmos]

Hi, am I blind or is the "Scroll to Bottom" missing on the mac app?

[urban schooner]

I thought this was the pro members server given that the link led me here. I've been raging about the lack of answers and support on the wrong server it seems.

How do I join the pro server then?

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1111786888626438245 try now

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1225557012532695060

[half tangle]

Hi ! What is the model used for Deep Research ?

[brisk breach]

Also Pro, can I get in too?

[viscid jackal]

hey, can someone explain the "deep research" function from perplexity? is it a conbined search of pro search and R1 reasoning?

[bold onyx]

reason -> search -> reason -> search in a loop

[viscid jackal]

yeah, i noticed that too. perplexity seems to perform a reasoning search routine, which argues every single search step. after that, it performs the regular R1 reasoning.

[manic gazelle]

i also have pro

[fringe acorn]

same here

[fringe acorn]

@ionic sparrow

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1111786888626438245 hop in

[quasi tusk]

can i paste the whole chapter of a book and ask him to summarize?

[rotund salmon]

Hey guys
Using API request and sonar model i receive incomplete responses. Maybe I need to receive this data after some time so that the AI ​​has time to respond completely? I have the ID of completion, but what url can I use to get it?
@ionic sparrow

[swift trout]

Are you enabling the stream option? That will basically stream the response word by word

[rotund salmon]

stream: false

[sick bear]

does anyone know what the daily voice searches are for pro users??

[vague orbit]

There is no consumption based cost unless you actively activate an API yourself, correct? Worst that can happen is that you hit your daily usage caps on the pro plan?

[sick bear]

Are the daily usage caps at 600 per day for the pro plan?

[haughty brook]

You have a 600 usage limit per day for pro models and 500 for reasoning. The default model has unlimited uses.

[haughty brook]

Correct, the api and the website are 2 different products. No extra costs for using the website if you hit the limit.

[serene nimbus]

Bro same here

[white elm]

i'm using the api but the results are terrible. even sonar-reasoning-pro the results are quite subpar compared to what i get with pro search (through the UI). any way to fix this?

[rich sage]

💡-> Tip: "Copy this sentence without adding anything else:"

Prompt:

Copy this sentence without adding anything else:"A futuristic industrial design render of a chair made of transparent orange acrylic with metal supports, featuring soft curves and glossy finishes. The chair is displayed on a matte gray backdrop with soft shadows."

[ionic sparrow]

#⚡│ask-community message

[vestal lintel]

[untold barn]

Hey, how does the AI Model option work? Starting a new conversation I'm choosing between auto, pro, r1, etc. so as I assume I'm choosing the model there. What does that setting do?

[inland musk]

is there a way to remove the "Related" section that is added to after the ai-answer? it is very distracting for me

[summer estuary]

this is your default model for every new conversation, so when you dont choose a specific in each conversation, this will be the main

[white elm]

unfortunately doesn't seem like there's a solution except doing it yourself. i'll just reimplment this with exa instead.... : (

[untold barn]

Thank you!

[ionic sparrow]

#👑│pro-lounge message

[stoic crest]

I tried claude 3.5 sonnet on perplexity and claude site using their 3.5 sonnet model (I have paid subscription). Response on perplexity was very poor comparing to claude. I tried again with new chats and same type of poor response from perplexity. I am wondering if perplexity is using cheap sort of model and saying it's actually claude 3.5 sonnet?

[tawny hatch]

Auto uses the model you choose in settings automatically. R1 and o3-mini use those models of course, and deep research uses r1 it seems to an extent.

Rewriting of course lets you choose the model you want to use for it

[tawny hatch]

Yeah I've noticed the same. Exactly why I don't use perplexity for writing or anything that doesn't require web search and sourcing.

Perhaps it's the system prompts and whole web search that makes it go a little wonky

[stoic crest]

I set 3.5 sonnet model from settings

so you can't disable web search?

[tawny hatch]

You can, but I still don't like it's answers within perplexity

[stoic crest]

I see

I think perplexity is playing with their pro users

It makes no sense to offer claude and other models so cheap

[onyx surge]

Could you please assist me in disabling the Follow-Up function? I find it difficult to highlight content, and it feels a bit inconvenient (web, Firefox)

[rich sage]

An extension (Complexity) allows you to do this: "Prevent headings from being rendered as follow-up links." (https://www.cplx.app/)

[south steppe]

How is spaces supposed to work with collaborators? I've invited a friend as collaborator and they have accepted my invite. However in the space we can't see each others threads unless we specifically share them (like you can do with any thread), aren't you supposed to see everything in a space by default?

[foggy bolt]

hello i have a quick question. Perplexity generated me an image - circle with 3 colors for my discord profile.
i tried to replicate it but with an added symbol that was asociated with a game i play in the middle. its refusing to generate anything for me now and says it cant generate images even though it did before. im also Pro.
any help would be apricciated. thanks in advance

[hollow ginkgo]

Hey @foggy bolt!

If you find the original message helpful, please consider reacting to it with the :star: emoji. If the post is appreciated by the community and receives 5 stars, it will go to the https://discord.com/channels/1047197230748151888/1082806833938436228 channel and the post author will get the <@&1082034222778302614> role on Perplexity.

[olive belfry]

Anyone on here use perplexity for sports betting research?? I am new to this but what questions do you ask if you don’t mind me asking. I ask a question like predict the score but that seems pretty basic. Thanks

[frosty wave]

Hey, is there any live tally of the IITs signups thing?

[crimson dirge]

I am new to Perplexity and have signed up for the deep research feature, which seems impressive. However, I’m struggling to write articles at the level of Claude. Can someone guide me on how to write a proper article with Perplexity? What am I missing here?

[wet ocean]

What are you trying to accomplish with article writing? There is a feature in Perplexity called Pages that you can use to focus on writing information articles.

[fading hazel]

Few questions when using pplx:

1. Does the context window of llm providers stays the same, I.e Claude 200k? R1?
2. Does pplx use cached tokens in supported models, e.g Claude?

[weary kindle]

are your deep research threads saved somewhere? i cant find it in the library tab.

[fringe acorn]

in ur regular threads library

[weary kindle]

just started a new one, and this one actually saved, but my old one didnt. strange

[stuck jolt]

Is there a way to install the app outside of App Store on macOS? The App Store is broken (I've seen several poeple reporting that issue) and I'm unable to install any app. Anyone has an installer?

[fluid crag]

What o3-mini model is used in Perplexity? low , medium or high????

[dense tundra]

i would assume its low

[grand junco]

what would be the most appropriate AI framework to match US Federal Laws with court opinions that rule on them in each circuit?

[stark sparrow]

Is there any way to use elevenlabs for direct output with perplexity? That is to say, I ask a question and perplexity responds with the elevenlabs voice that I choose.

[fluid crag]

but what? low or high?

i know that reasoning with o3-mini but model low , medium , or high?

answer?

[ionic sparrow]

https://www.perplexity.ai/help-center/en/articles/10354919-what-advanced-ai-models-are-included-in-a-perplexity-pro-subscription

[fluid crag]

o3-mini high ?

The most up to date reasoning model by Open AI. XD but model o3-mini high?

support didn't answer me in this question

[ionic sparrow]

#⚡│ask-community message

[crimson dirge]

Thanks for your reply. I just wanted to use the deep research and write a blog article based on that. The research looks great, but I just can't seem to find an easy way to prompt to write a comprehensive article.

[wet ocean]

Ah, so what you might be able to do is do a deep research prompt, and at the top under your prompt should be an export button (see image). That most of the time has the option to Export as a page.

In the Perplexity Page, you can continue to expand upon each section by having it rewrite it using Detailed instead of Concise

Or, if you prefer, you can simply Export it into Markdown for your own Blog

[regal field]

I often mistake this for an actual link leading to an external site, and you know what the worst part is? You can't even select/copy the text -_- (firefox browsers only)

[hexed robin]

Does anyone know why if you use Perplexity without a browser extension like Complexity, you don't get access to several different models like Sonar XL or Haiku, but if you do use it you do? And then follow-up, how come Perplexity "hides" this, as, when you use Complexity on one computer, you can change the specific model, and then when going to a different computer, you can see that the model is actually present in working?

[summer estuary]

Will Chat GPT 5 added to Perplexity?

[scenic shell]

Can I configure these options for https://www.perplexity.ai/search? I want to know ther is an documentation about this.

[next wyvern]

Is there a perplexity Google new tab page?

[vapid maple]

it is broken currently, only keyword q= works

[manic onyx]

will grok 3 be added to perplexity?

[scenic shell]

OK Thx ;)

[left ether]

chat gpt 5 aint even out yet what are you talking about lmao

[lament cypress]

I chose to cancel Pro subscription only one day after subscribing it. Why does it show "You're not eligitable for a refund." According to the refund policy https://www.perplexity.ai/hub/faq/billing-and-subscription, I confirmed one-day trial fulfills the refund policy. I reported my case to Pro support via Intercom and emails, but no useful information is provided. Does any one have idea about my case?

[void oracle]

can i ask, im new on perplexity ai pro plan, where is wolfram to use for math questions?

[dense atlas]

hello everyone! Just a quick question: I've opened a ticket through email support last saturday. How long usually does it takes to have an answer?

[ionic sparrow]

a few working days

[dense atlas]

thanks! hopefully soon 😄 I can't wait to have my pro subscription 😄

[ionic sparrow]

let’s get you properly prepared for that PRO account

deep dive ➤ #💬│general message

[dense atlas]

thanks for sharing!

[summer estuary]

bro i said "will", ik thats not released yet, but when its released then will it be added?!

[wet ocean]

too many variables to determine yes or no

they do try to implement new models when they come available, though

[fringe acorn]

@summer estuaryAre u Thomas Shelby

[summer estuary]

yes i am

[olive gate]

Hi! Is it normal that modes utilizing R1 in some capacity (deep reason, reasoning with r1) started looping into repeatedly iterating "but wait", "but how", "wait" etc. ? This started happening somewhere in the last 48 hours, and leads to absurdly long chains of thought.

[crimson dirge]

That's a great idea, thank you. I'll play with that.

[brazen coyote]

Hi, where is filter option for math, writing etc in browser?

[void falcon]

I'm confused, I use perplexity on Android and I see in settings an option to choose another AI model. currently I'm using something named Sonar but I also see Claude, gpt4-o, grok-2, Gemini 2 flash, auto. why would I choose another AI model? does this mean that the perplexity app can replace the other AI apps on my phone? if I set to Auto can I somehow specify which model I want my answer from?
thank you

[jovial quarry]

hi guys, is there a way to see how many pro/deep research searches I have left on my daily limit for a pro account?

[terse crystal]

Hi I just suscribed to the pro version and I would like to know if I use deepseek R1 model can I put more files than in deepseek because I mainly suscribed to the pro version because of that. I would like to use deepseek model but I’m very limited for the length of the files I need to dl to the chatbot

And they say that the limitation is up to 4 files but it’s not enough for the pro version little bit disappointed

[winged badge]

there is a chrome extension "complexity" it keeps a track whatever we search only on browser tho

[main oxide]

Hi, how do I get in touch with these folks? I need an enterprise level demo and it’s so hard to get in touch with Perplexity folks. Can someone please direct me. I have already emailed them at their enterprise email address as well as their support email but I’ve not heard back. It’s been two weeks and it’s extremely frustrating to get in touch with them. Can someone please connect me with the right person?

@hollow ginkgo

[dim canopy]

Hello, how to générate image with dall e in the Android app please ? I have the pro version

[topaz elk]

Hey perplexity in assistant mode - mic seems disabled, app has mix permissions enabled - android Google pixel 9

[rustic willow]

can i upload 19 studies and have it run similar characteristics on all 19 studies with pro account. if so, how?

[haughty brook]

Generating images is currently only available via the web version

[jagged mountain]

Hello does anyone know why i cannot use focus mode anymore on web perplexity? I can still use those on ios perplexity

[summer estuary]

in the main settings you choose you default model. this will generate any future answer. if you want another model for an specific question then you must choose it at the thread while typing the question/ prompt.

Thread

[fluid lark]

Can someone give me the URL parameter to toggle Deep Research? like copilot=true is to enable Pro Search?

[summer estuary]

main in settings @void falcon

[buoyant wasp]

Hello can someone share interesting use cases that use Deep Research?

[grim vector]

Is there an issue uploading files to spaces?

[dawn linden]

What model do people use for coding? I use R

[late pecan]

How do I change my email address may I have a pro account to another one cuz I have been trying to Pro account accidentally into the force email address I bet you said I was locked in with the voice and it recognize to change

[haughty brook]

Please email [e-mail] they can help you move the subscription to another account

[lime geode]

How do I get the most non-trivial answer from Perp when asking for creative things like image generation prompts or so (eg. to avoid boring AI common default ideas that AI falbacks to and became obvious to humans) or do you have any custom instructions to humanise the thinking model even more?

[ionic sparrow]

#👑│pro-lounge message
#👑│pro-lounge message

[sullen marlin]

does your perplexity also recently timing out requests? I noticed it happening often, mostly with geopolitical questions with deep research but this might be just coincidence. Obliviously I am not going to share question

[topaz elk]

Bumping this

[summer estuary]

is the perplexity page that slow for you too?

[hollow nova]

Try uninstalling and reinstalling

[onyx prawn]

xAI’s Grok-3 when?

[obtuse surge]

when will be grok 3 added?

[manic onyx]

grok 3 ⁉️⁉️⁉️

oh it doesn’t look like grok 3 API has even been released so need to wait for that

[topaz elk]

Didn't work

[regal turret]

My assistant's mic doesn't work in bank apps, because of the phone's security measures.

It can be your case too.

[half cairn]

Someone tell me sum Abt this lol

[manic briar]

Will perplexity eventually have a grok-3 model for use

[regal turret]

Grok-3 doesn't have API yet.

[crude terrace]

I noticed an article mentioning that PPLX has open-sourced the uncensored Deepseek-R1-1776 model and indicated its availability through the Sonar API. However, I can't find this model listed in Sonar API's Supported Models documentation.

Any idea why? Is there a delay in updating model in Sonar API?

[cosmic summit]

I want to subscribe to a service, but my card gets declined at the registration stage. Why is this happening?

[paper isle]

Is Perplexity capable of reading a 795-page PDF? Whenever I upload it, it says it has read the file, but it only processes the first few pages. I have some questions, and they are based on the content of the PDF. While answering, it needs to search through the entire document to find the relevant information. What is the best way to do this? I am using Deep Search.

[fluid lark]

Use the Gemini model, it has a much larger context window than other models and can read bigger files.

[paper isle]

Yeah got it, thank you very much for the info

Whether we need to keep it auto or we need to choose like pro or deep research in chat?

[terse apex]

Is there any way to use Deepseek R1 on perplexity via the API?

R1-1776 would be fine too.

[fluid crag]

how to get free credits in Perplexity?

API

[haughty brook]

You get free 5$ credits each month when you have an active pro subscription

[shrewd knoll]

Report to @slim night to get free $5

[worldly turret]

i would like to ask support for subscription, can i do it here?

[pure cargo]

How can I contact a human at perplexity about perplexity enterprise? I submitted a contact form and haven't heard back in about a week

[slim night]

Hey @pure cargo! Have you sent all the details to our [e-mail] email?

[pure cargo]

No I have not, I just submitted the form. I'll send an email. Looking forward to hearing back from your team

[manic onyx]

how does perplexity make money... like they provide more usage of o3 mini than chatgpt does for the $20/month sub

[late cosmos]

o3 mini high is quite good but o3 mini "classic" is meh in my opinion.

[midnight cloud]

How does enterprise billing work? I see can buy credits, and I can also auto-top up. But I seem to have a running invoice beyond what my credits pay for. Am I running up a monthly tab that gets invoiced at the end of the month? If so, why bother with credits?

[river ruin]

Model choice is a mess in PPX UI. You get a list of models in Settings ( Sonet/Sonar/4o/Gemini2.0 Flash, Grok2), and a different list of models to pick for a new thread (Pro/Deep Research/R1/o3 mini), with no overlap between the 2 lists. How are you supposed to make sense of this?

[manic onyx]

use complexity extension

[midnight cloud]

I'm asking about API billing specifically. Not the per-seat billing

[dry forum]

Does any one Know how to use Perplexity deep research through API ?

[wooden relic]

https://x.com/AravSrinivas/status/1891905511286768018

Aravind Srinivas (@AravSrinivas) on X

Congratulations to @xai for building world-class reasoning models in such a short period. This is a super strong response from America to DeepSeek. America builds fast, and xAI team and Elon are setting the bar for it! When the API becomes available, we'll integrate!

X · Feb 18, 2025, 5:39 PM

[serene stratus]

Hi, I use macbook and it's connected to monitor. The font size in the Perplexity desktop app cannot be changed to larger one and the texts are just too small, making the app unusable. can someone help me?

[sweet ledge]

I’ve created my own space and research thread on Perplexity AI, but I’m running into an issue. Whenever I try to load my thread, it takes a really long time for my answers to appear, and I constantly have to refresh the page. My thread has gotten quite long, but I’m hesitant to delete it because I’m worried it will erase all the valuable information we’ve discussed so far.

Has anyone experienced something like this? Any tips or suggestions on how I can improve the performance or prevent this from happening? Thanks in advance for any help!

[regal turret]

Complexity extension sometimes causes performance issues in long threads.

However, you can ask Gemini (for long context) to summarize all the valuable info.

Then just insert the doc in the new thread or even space itself and continue the research)

[prime orchid]

this Complexity feature has to be implemented to Perplexity. I agree with @river ruin . It does not make sense the model choice in PPX UI

[umbral forge]

when grok3 will be coming to perplexity?

[fallen flicker]

Hi, on android, paired and connected to BT earbuds, the assistant only listens to the mobile phone's mic. any workaround? Output is through BT.

[wraith vale]

What's going on here.. no one seems to answer people's questions?,,

[regal turret]

Discord is a chill place to hangout and talk about various topics. If someone really needs help — Perplexity has support through email.

To solve problems you need sensitive info. You know, info you should never share with people on Discord)

[wraith vale]

There is a user above who asked a simple question such as which AI program do I use I have seen zero answer I am Generation X failure is not an option to me that is failure by whoever's moderating this and yes I'm a pro user

[regal turret]

Zero answers... yet. We just need to wait in here. It's not support — just a Discord server)

[wraith vale]

Okay not directed to you just an overall View I step outside and I look in and that's what I'm saying

Have yourself a nice day

Thank you

[regal turret]

You too, brother))

[runic lake]

Does $5 API Credit Mean you still have to enter your credit card information

[crude beacon]

No

It'll automatically charge your pre-approved stripe payment method (you can setup so that it charges other card too ig)

(one you used to purchase pro)

[sullen harness]

What you use for web programming? Will it be as good as paid chat gpt? Claude or something else?

[dense tundra]

i personally use claude and o3, worked the best for me so far

[sullen harness]

Thx will try tommorow 😊

[astral sonnet]

I'm in tech sales. Here's the goal:

I have a CSV with 230 businesses and their websites. I need to rate them all 1-4. I'd like to input this CSV or paste it and have the model search every business on the web and rate the business based on my metrics (don't worry about those).

I've been trying Perplexity with R1 and then o3-mini since I heard it hallucinates less, but the actual output seems limited. It seems to ingest the CSV, but i can only output like 20-30 accounts, line by line. I can't get it to output all 230. I realize this is a big ask to basically have AI research every account individually based on 1 query and framework, but I'm curious is there's a better tool for the job here.

[crude beacon]

I'm not sure why you're not batching the input?

If you batch 10 businesses in a query, you can get your data values in 23 queries/prompts?

[astral sonnet]

i could, but could likely just do it faster mannually 1 by 1 at that rate

[crude beacon]

Aren't you doing this through API?

[astral sonnet]

no, just in a perplexity space using o3 mini

[dense tundra]

i would set up a script simply that batches this, puts it into perplexity each batch, saves the answers to a file and done

[crude beacon]

Me too ^ ^

[dim dagger]

What is the diferent perplexity ai and google search engin?

[astral sonnet]

is there a tool that can handle a request like this at scale? 200-600 accounts, for example? I wouldn't care if it takes a long time

[crude beacon]

Try setting your model in the settings to Claude

Claude has higher tokens output and input

it could probably solve your issue

(probably need to use pro search instead of r1 and o3?)

[dense tundra]

from what i know most models are like capped at 500 lines roughly, atleast thats when it always cutoff for me roughly

[astral sonnet]

i'll give it a quick try with claude in the space + pro search

[crude beacon]

Mostly these data are first stored in a large search database such as BigQuery and are batched by BigQuery (for any LLM task)

[astral sonnet]

pro search is built on R1 though i believe

[crude beacon]

Try to use auto or something ig.

[craggy thorn]

Hi everyone, I've a question about API here.

First of all if I'm in the wrong chat of course tell me !

I know that it usually is a common problem with LLMs APIs but do you see a big difference between the API and the web completions ?

Cause indeed it's normal for some things like Openai and we know well what the differences are, but on Perplexity I just find it weird cause I've realized that Perplexity litterally ignores some of the sources we give to it, or at least does not really follow them.

I had a prompt that was working perfectly on the web where I asked perplexity to go through a few pages of a website to analyse things, but once I started using it on the API I had to fine tune it a lot as the correct realization of the "scraping" or so (I don't know the exact way perplexity handles website searches) it was much more sensitive and sometimes it was not even following my link, and then gave wrong analysis.

Does any of you know something about it ?

Thanks in advance for your answers 🙏🏻

[topaz elk]

Microphone doesn't work in Android 15 for assistant

[fluid lark]

Hello I ordered clothes from perplexity but I wanna return it’s too small what should I do

[umbral lily]

Hey guys, I'm kinda new to Perplexity, I was just wondering, on the Pro version what does it mean we can use Deepseek R1, OpenAI o3-mini, Claude 3.5 Sonnet, Sonar, and more? So we pay $20 and get more than one AI model? I'm confused 😅

[summer estuary]

yes, right. You can choose different models and switch between them. You also have 2 reasoning models and deep research

[vapid maple]

yes

||pplx is more or less about ai powered web search, the reply length is capped||

[summer estuary]

you have gemini 2.0 flash, grok 2, sonar, claude 3.5 sonnet, grok 2, gpt 4o

and for image generation: Playground v3, DALL-E 3 and Flux.1

also higher usagelimit for deep research

[umbral lily]

Wait, that's so much worth it wow, so basically all the main "pro" versions

[summer estuary]

yes

and in a few days/ weeks, when xai released theyre api. the new grok 3 too

[umbral lily]

But just for searching, or I can use the main capabilities of each one (like data analysis with deepseek and image generation with Dall-E)?

[summer estuary]

you can use every model for its own

you can create spaces like i do, and choose one specific model for each space

[vapid maple]

these models are avail in pro

[summer estuary]

in the settings, you will set your default image generation model and text generation model

[umbral lily]

damn, sounds way better than paying $20 for each one of them

[summer estuary]

yes, and its getting more

[umbral lily]

I see wow, I'm gonna give it a thought honestly, that sounds crazy nice

Thank you guys!

[summer estuary]

you can also create your own pages and publish ai generated biographies and more, but its only necessary if you use it

[summer estuary]

you can get it at first month for $10 instead of $20

i can send you a link

[smoky otter]

Hey everyone,

My question is about the Pro features and how reasoning models work. In settings, we choose a default model (like Claude 3.5 Sonnet, R1, etc), right? Then there's that blue toggle for additional reasoning models (R1, o3-mini, etc.). Here's my confusion: does this toggle combine the chosen default model with a reasoning layer on top, or does it completely replace the default when activated?

Essentially, is it a two-step process where you first select your base model and then add reasoning enhancements, or are they mutually exclusive settings?

[summer estuary]

no i think it will replace it for that one search or generation

[smoky otter]

thanks!

[smoky otter]

Well, if I'm already here, I have another question.

I've noticed two separate issues:
First, workspace default prompts don't seem to work at all. For me, it's mostly about languages that it should use to search information and to answer me.
Second, there's been an interesting evolution in how Perplexity handles language for searches. Before R1, it couldn't automatically determine when to search in English for questions asked in my native language - this had to be explicitly prompted. When R1 was released, it showed some improvement in automatically choosing English for searches even when questions were asked in my native language (though it would sometimes still reply in English rather than my native language, which wasn't ideal but was still better as the answers were more comprehensive). However, after some recent updates - either server-side or in certain clients - R1 seems to have lost this ability and now most of the time cannot do it without explicit instructions again, like the other models do.

It bothers me to specify it every time, and I often forget to do it and have to edit the question. I even created a Raycast snippet, like: "search in English, answer me in my native language".

[umbral lily]

Oh, I appreciate it man, i intend to sign something later this year, it's been real useful to me, but I'm kinda short on money lol

[night perch]

Can’t seem to find the button to publish pages. I have pro subscription from Revolut Metal. I wonder if it’s a limitation. Have you encountered the same?

[full current]

Hello everyone, I’m wondering, with my pro version, I have several models to choose from, but do these models apply only for auto, or are they also applicable to both deepreach and pro?

[ionic sparrow]

#👑│pro-lounge message
#👑│pro-lounge message

[echo ether]

Our company is considering a corporate contract with Perplexity.
We are unable to use Deep Seek's AI model due to compliance issues.
So, I have a question. When signing a corporate contract, is there a function to restrict the AI ​​models that members can use?

[modest ledge]

Why does o3 mini in perplexity not respond with more than 13k characters? It doesn't go above 2000 words response.
In chatgpt free it responds with over 5000 words in o3 reason when asked

[fluid lark]

I think it's because the response has a character limit in Pplx

[ionic sparrow]

please contact [e-mail]

[fossil tartan]

is AI Data Retention off when using incognito mode?

[brittle axle]

I have pro subscription and trying to use deep research tool but it automatically turns to pro. It was working fine 2 days ago... is there any issue currently?

[fair valve]

Is there a way to use my chatgptPRO subscription with Perplexity Pro? I hate having them both open.

[haughty brook]

This is not possible

[reef ginkgo]

Does perplexity have an API?

[regal turret]

#🧪│api-general

[ionic sparrow]

https://www.perplexity.ai/hub/blog/introducing-the-sonar-pro-api

[fair valve]

Thank you. Never hurts to ask. 1.0 reasoning is so good right now.

[lament swan]

Does perplexity have any functionality similar to chatgpt operator? if not, is it planned?

[copper jetty]

What version of GPT-4o does Perplexity use? Is it [chatgpt-4o-latest-20250129], or is it an earlier version?

[covert edge]

Hey, I have a dozen perplexity accounts and api keys and from time to time a key stops working entirely and i have to go and create a new one and in the dashboard it shows the key that is getting the 401 error as being grayed out. Any reason why this happens and what I can do it fix them or prevent it from happening again? This is my 4th time having to regen a key this year.

[hasty charm]

It does not appear to be the latest, since the knowledge cutoff date was updated in the new 4o release to sometime in 2024 you can check with a query about that:

[wispy dawn]

Is there a way to have perplexity crawl a website by providing the url in the prompt?

When I provide a site URL and ask a question - perplexity just straight up lies in its answers, but maybe im not using the right syntax?

[bright ridge]

Just got an email from Perplexity with the subject line: "We've canceled your Pro Subscription". WTF? Then in the body of the mail it says: "Hi there,

We noticed you canceled your Perplexity Pro subscription and wanted to let you know we've downgraded your account accordingly.

We understand there are many reasons you might have decided to step away from Pro. We would love to hear your thoughts on how we could improve Perplexity for both you and other members. Your feedback is invaluable as we work to create the best possible experience." This is weird and I didn't cancel anything. Can someone please help?

[balmy kindle]

for some reason it seems like Pro deep research searches are lower quality than the free ones, what could be the problem?

[viral cipher]

I subscribed through Bouygues to Pro and cannot see image generation models..

[rich sage]

In your settings/account https://www.perplexity.ai/settings/account (Only web broswer)

[mint imp]

how do you even generate images?

[rich sage]

https://discord.com/channels/1047197230748151888/1225557012532695060 or #⚡│ask-community message

[upper crown]

Is it possible to wind the text to speech speed faster in the android app?
It's awesome, but I'm used to fast speech - I wind my podcasts at least to 2.5x speed...

[steep horizon]

Is there any way to personalize the look of my question? make the font smaller etc, it looks ugly as it is disproportionately large compared to the answer. Overall the chat interface is not very user friendly

[rich sage]

Not directly in Perplexity (unless they decide to change this font size), but an extension called "Complexity" (https://www.cplx.app/) allows it. Among other features, it includes an option to edit the theme and use your own CSS tags.

Here is my CSS, for example, for the font size and line spacing of the query:

  font-size: 1.05rem;
  line-height: 1.3;
}```

[wet ocean]

neat, where did you add that?

[rich sage]

In the Complexity dashboard: Themes / Local Themes, the small pencil icon to edit.

[wet ocean]

ah, didn't even notice that there, very cool

ty

[proud quartz]

Hej, czy ktoś wie jak można ograniczyć na stałe perplexity tak jak to można zrobić gpt do tego, oczywiście nie wpisując testu za każdym razem: jesteś w 100% szczery w swoich odpowiedziach nie będziesz wymyślał żadnych faktów informacji ani danych. udzielasz informacji gdy jesteś czegoś pewien w 100%. jeżeli czegoś nie wiesz lub nie jesteś pewien to mnie o tym poinformujesz.

[sinful path]

Hi, does anyone know when the perplexity business fellowship starts? if there's an agenda?

[serene ridge]

Hey little question is it normal. When I say on iPhone Siri Perplexity. Siri Only Show the Text and not Read the Text, Siri Voice Settings all right an on

[trim tide]

How do I know which model is Perplexity using? I kinda think is pointless that we are able to set a default model in settings but during an actual query there’s no indicator of which model I’m using. I keep hitting 🔁 so it regenerates the answer with the model I prefer. I also remember being able to disable pro search and now I can’t do that anymore.

[wet ocean]

it tells you in the bottom right corner of the response

also, use Complexity plugin

[echo aurora]

Hi, does the latest "Sonar" model has online search functionality? I'm using the "llama-3.1-sonar-large-128k-online" and would like to update into this model.

[glad nest]

Yes it does have web access

[echo aurora]

Thanks much!

[leaden spoke]

How can I use Claude 3.5 Sonnet

[leaden spoke]

I have set it up in the setting but it won't show up

[hasty charm]

This is a pretty common question because it's not very intuitive. After selected, pro searches should use it, not sure about auto. Additionally, each space created has a default model you can set. You can also verify the model that should have been used to reply from the lower-right corner on a response (hovering over an icon will show the model).

If you notice any issues, let us know!

[indigo jasper]

Hi i have a Problem i Need a quick change of my Pro Account email Adress. How I can do it? Because I need perplexity on windows. But I use a privat apple relay mail adress, but doesn’t work.

[haughty brook]

Please send a mail to [e-mail] or send a message via the intercom. They will be able to move your subscription to a new account

[haughty brook]

#👑│pro-lounge

[leaden spoke]

Thanks! got it now

[harsh matrix]

Hi - I've checked the message history and haven't seen anyone answer this question yet. I upgraded to Pro, because it mentioned that it allows unlimited image uploads. I still can't upload more than 4 images. I have 10, so it's not a crazy amount. I just need any model to extract some titles from the images. Anyone know a fix to this?

[untold barn]

There's no image uploads limitation, but there's upload limit for files per message (or thread, I'm not sure).

[hasty charm]

I read somewhere that images should be less than 25 MB... but that's such a large size for an image already that it probably doesn't matter.

[tropic storm]

Hi, how does rate limit work on Perplexity API? Is it per account or per key?

[latent zenith]

Is there a way to save assistant sessions into history?

[fluid lark]

Why does Pro Search fetch 50 sources but only cite the first 10 of them?

[plucky vortex]

Where i can write promotion code ?

[regal turret]

It's already in the history.

[neat olive]

I tried it for hours when I intended to create something like a fact-checking prompt, and I came only to hallucinated results. In discord I found no solution. I believe it provides more accurate answers if you upload the text as a file and ask your questions about its contents. But, I didn't test it.

[glad nest]

What is the max output token in Perplexity AI?

It seems to be inconsistent.

[glad nest]

There is no code. You must visit the referral link

[dense tundra]

does anyone know how to fix the lag in long threads? loading the thread up takes for me half a minute minimum atm

[dense dirge]

hello, I'm not sure if this is an issue or a known thing, when I'm using perplexity in web browser and have a longer thread which I would like to continue because of existing context, what's the best practice to do that?
the issue I have is that the thread has become very very slow and unresponsive as soon as it got a bit longer

[fluid lark]

Will Claude 3.7 Sonnet be added?

anyone know when grok 3 is coming to perplexity pro

[wet ocean]

when an api for it is released

[fluid lark]

Will Claude 3.7 sonnet be released

There is already API for it

[stiff prawn]

So, I used Lexi (Perplexity) to assist me in writing a new master resume. However, when I attempt to get her to assist me in trimming that down to the best fit for a job posting that I give to her it seems that she isn’t looking at my entire resume. I try to copy and paste as plain text, but then a paste.txt shows up on the screen and she only uses the first page or two in writing the job specific resume.

Is there a buffer size limit that I am running into inadvertently? Is there a different way I should be loading the information into her? I am using the free version at the moment because I wanted to see how it worked before investing the few dollars I have in my bank account on the paid version. Thanks for the assist.

[feral rivet]

Maybe you are running into an token output limit, you can just say continue where you left off, then you can stitch it together in a text editor

[stiff prawn]

Well I am trying to input not output and after asking her apparently there is a 20,000 character input limit.

[fiery lark]

Hey how could I reach out to perplexity business development team? Thanks 🙂

[silver quail]

Hi there,
I’m a tech journalist specializing in AI-related articles for "Les Numériques" (a French online media). How can I get in touch with the Perplexity team? Thanks 🙂

[glad nest]

yes, it's so laggy when using Deep Research. I suspect because the content is dump directly to the browser.

[fluid lark]

Btw What are the differences between claude 3.5 and 3.7

[native silo]

Sorry if this is not allowed.
I'll delete this post.

I have an year Pro subscription (promo code) that I don't need.

I have Pro, and everyone I know also has it, so if anyone needs Pro, please send me a DM and make me an offer.

It would be a shame if it expired without being used.

[sullen nebula]

Anyone know how to get 3.7 claude enabled?

(I have pro)

[regal turret]

Hello. You've added 3.7 Sonnet. Arigato.

Any chance on getting It's thinking version?

[regal turret]

Only on Web rn.

[gleaming viper]

i subscribed using the app on ios / iphone. How do I download my invoice?

[haughty brook]

Go to the appstore > click on your pfp in the top right > purchase history. There you can download your receipt.
Or go to https://reportaproblem.apple.com/ on web.

[kind tulip]

how do u guys use claude on perplexity

cant seem to find the model

[haughty brook]

You can find the models on your settings page https://www.perplexity.ai/settings/account

[kind tulip]

oh ok i fell in love with claude

do u know if its the exact same output

and what is this for, if i chose the ai model in the settings

[haughty brook]

The output won't be the exact same as on claude website. Because of the different system prompt. that dropdown is for reasoning models/ pro search. If you choose Pro/Auto it will just use the model you've chosen on your settings page.

[kind tulip]

thank bro ❤️

[gleaming viper]

Can’t see any receipt download option

[gleaming viper]

Maybe I need to wait a while ?

[static token]

The letter about the waitlist says the following: "Want access sooner? Share Comet on social media and tag our account." Does this mean that it will generally speed up the distribution of access to Comet for everyone, or just for me?

If it’s just for me, how can I get noticed? Tagging the account is only half the job, and it’s harder for those handling access to Comet to find me in that case. Sorry, it’s just that the text of the letter isn’t entirely clear.

[wicked summit]

For deep research what AI model does perplexity use?

[keen venture]

Hey guys. I have chosen the AI model as Claude in the settings but it almost always reverts to using GPT 4 Omni. Is it because I have files attached or why is that?

[neat ivy]

i have a perplexity pro account do i need to buy the api keys for development?

[serene oxide]

hey what models does the Pro search use ?

[fair marsh]

Every extensive prompt that I send through the Perplexity API using the sonar-reasoning-pro model is limited to a maximum of 10 citations/sources. I want the response to use more than this. How can I achieve this?

[prime hare]

How are colloborative spaces supposed to work like can the other person see the threads?

[random shadow]

Am I the only one for which the website is SUPER slow (like when I type a letter there is a 1s delay) and even crash very often ? My laptop is not cutting-edge but should be sufficient to run a website (16gB ram, i7, ...)

[pale crown]

Hi there 🙂
Is the copilot function currently deactivated? Can't find the toggle on web version or android app (Perplexity Pro account from Germany)

[prime hare]

Whats that?

[opal robin]

Has anyone been able to put together helpful prompts to use for medical / health condition search? Dealing with chronic illness that doctors can't help. Thanks in advance

[rustic wing]

Hi guys new here.
Have been exploring platforms like perplexity and other such aggregator.
Can anyone share there view of good perplexity is compared to other such aggregator?? Or if they can point out where can i look it up..
You.com is another platform that looks good

[pure cove]

Where in the app or on the browser can I find payment receipts or invoices?

[pale crown]

Perplexity Copilot was announced (according to their youtube channel 1 year ago) https://www.youtube.com/watch?v=ht3XV_nbduQ

I really would like to use that feautre. But it's just not there... 🤔

[astral oyster]

Copilot is just pro mode now but it was only the name that was changed. Pro mode does more now with all the models than before.

[pale crown]

Thx for the answer 🙂

[coral blade]

Is it a good idea to replace google search with perplexity?

[rustic wing]

whats the context size on perplexity ? and what's the daily or monthly limit of the queries ? asking for perplexity pro

[prime hare]

only if you are paying for it

makes you feel you are getting your moneys worth

[serene stratus]

hi, can we restore deleted thread in perplexity?

[stray current]

Is the perplexity dev meet up free to go?

[ionic sparrow]

please contact [e-mail]

[gray rune]

yes~

no cost to attend

[stray current]

Ohh thanks I have requested for the pass l. Let’s see if I get it…

Also, is there a age restriction like only 18 or above?

[gray rune]

if you're a young dev based on SF who is building with Sonar, we'd love to meet you!

[stray current]

Thanks

[tough sail]

how to access plaground v3

[regal turret]

Settings
Then create thread like usual and click Generate Image.

[dense tundra]

does anyone know how to generate images on phone using the app? i looked for like 15 minutes but couldnt find it 😭

[silent dew]

@gray rune I'm dying here and need support I'm not getting from intercom or the website.

[feral rivet]

You can try scite AI, you get a few free messages a day, but overall I don’t think AI is the best in terms of medical yet, a lot of slop on the internet that it will regurgitate.

[feral rivet]

You can’t, only via web

[dense tundra]

oh okay, thanks

[sand stump]

There must be a limit to the thread conversation length since, in a long conversation, my earliest questions and answers have been cut or deleted. How do I export the entire thread rather than save each answer one by one? I note also that if I use something like a cam scanner app to save the Screen, I can see footnotes but they do not appear in the resulting document and obviously they are not clickable

[eternal patrol]

Why can't I switch freely between models like 4o, Sonnet 3.7, among others? It only gives me this "Pro" search option, and I don't even know which model it is using in full. I would like to choose the model freely, but I am unable to do so.

[thick agate]

Go to settings and choose ur preffered model

[eternal patrol]

Ok but, It'd be better If It was possible to do It in a easier way, or If i could see which model It is using

[strong zinc]

Does everyone see this font behavior when you use Perplexity?

[vapid maple]

#🌐│feedback-web message

[strong zinc]

So it's normal? I don't know if I asked the AI clearly enough.

[brisk breach]

I prefer it. Just have to know to verify sources. But that's not much different than Google search

[feral rivet]

There is a community extension that helps: #💬│general message

[shy pivot]

What are the time limits of using Perplexity's new voice mode for Pro and Free users?

(rant - sorry: why do most AI companies get billions in investment but convey information about their products in such an uncoordinated and scattered way between social and outdated FAQs?)

[fierce stream]

I have perplexity pro via my school and want to use some of the reasoning models with a perplexity space, how many uses of deepseek R1 or o3-mini reasoning models will I be able to use?

[pale crown]

Is there a place to check the upcoming features? Like a product roadmap?

[fluid lark]

How many can you make in 24 hours?

[quartz schooner]

Something similar has been asked before without response, so I'll try. . . What's the character limit for custom instructions in Spaces? Want to find out to see if the feature can function like a Custom GPT. Edit - Looks relatively low around 4000 characters, bit disappointing.

[fluid lark]

Are there any plans to ever add Mercury Coder from Inception Labs to Pro? (https://www.inceptionlabs.ai/news)
As far as I know, it's the fastest AI at the moment

[formal tapir]

I am sharing a space with another pro account and he can use it but we dont see each other threads.
Is there a way to share all threads of several users of the same space ?

[fierce stream]

yes this

[tall elm]

hello, has anyone found a way to automate searching or asking questions against a document in perplexity? i am trying to generate summaries and other things out of transcripts and its tedious to do it through the UI one by one

[final oar]

how do you get the AI to stick to a story. I ask it to add something and I get people running into a gust of wind blowing into the ahllway and other stuff that's not relevant. its trying to add back gorund I suppose but its annoying and off topic.

[ruby saffron]

How can I switch from monthly plan to annual plan in android?

It only allows me to resume subscription two months after cancellation

[glad nest]

Has GPT 4.5 in Perplexity, yet?

[ember charm]

expensive..

[warm stratus]

Hello anyone manage to get images url from the perplexity API ? It returns only placeholder https://example.com/xxxx.jpg
And not a valid url, it's from a property website

[oblique thistle]

I have NO idea why they do this… been a pro member for over a year and this is my biggest complaint.

There’s literally chrome extensions at this point to do this…

There’s a work around and that is to start a chat, burn a prompt quota and then rewrite with the model you want.

But how’s this a user flow with USER in mind?!

To pour salt on the injury, you can now do this with “reasoning” models, but still not with the classic ones. Why??!

[e-mail]

[iron sparrow]

Does anyone know who at Perplexity AI I can get ahold of to learn more about the beta advertising product(s)? I have several clients interested, but nobody is responding on social media. Thanks in advance!

[kind tulip]

sometimes it codes in liek a weird layout, like textformat, i sthere a way to fix this

[eager minnow]

Hey, quick question - Can I refine a Perplexity prompt to search only a particular website/resource? As an example, I’m doing UX research on Typeforms and using Typeform’s Blog to skim through information, but I want to automate the process. Any tips would be appreciated!

[spring gyro]

Is anyone aware what is the context limit of 3.7? Is it same 32k as always, or was it updated?

[eternal patrol]

Vai ter Grok 3?

[sage granite]

The new voice interface is shockingly sh*t. I asked them to switch to an Irish or English accent and let’s just say…Dick van Dike did a better job in Mary Poppins…it was that bad. Apart from that perplexity cut our conversation off mid stream on 2 occasions. #TotallyUnderWhelmed

[wanton skiff]

edeem perplexity

[buoyant ridge]

Hello everyone. A quick question:
Does Perplexity's "Deep Research" AI service support each of Traditional Chinese and Simplified Chinese inputs and outputs?
(I am a citizen of and a resident in Taiwan, and am very very very appreciate this "Deep Research" AI service openly provided by Perplexity.)

[chilly rivet]

Why does Claude 3.7 sonnet have a skill base only up to 2023, like Claude 3.5 sonnet, when Claude 3.7 sonnet should have it up to October 2024?

[glad nest]

I'm just happy that now the prompt font is not taking up half of my screen.

[crude onyx]

Asking an AI model about itself usually leads to hallucination and confusion as the data it was trained on was before it existed

Anthropic say it was 2024 so just trust that as the correct information

[jaunty plume]

Hey quick question please - I am wondering how I can choose a specific model, like Grok or Claude. I can see them in settings but not in the available models in the search/chat bar. thanks 🙂

[cold vortex]

Any one facing issues with Chatgpt4.5 model or Claude recent model ? why it taking so much time almost more than 5 to 10 min . Even Sourcec section also not updating..

[regal turret]

Choose it in Settings
Use Pro
Done

Auto — fastest, but not smartest model
Pro — your chosen model
Deep — big pages with finetuned R1
Reasoning R1 — smartest boy out there
Reasoning O3-mini — faster than R1, but dry and sometimes stupid

[steady granite]

So which Model would be best for quick Wordpress snippets?

[jaunty plume]

Thank you so much for your help. I am also seeing an image generator model but not sure how to use it.. any idea?

[regal turret]

You should write your promt and click here:

Model will tell you, it can't generate images, when your query is done, but after clicking on Generate image — you'll see.

[regal turret]

I'd choose 3.7 Sonnet. It's just the best.

[jaunty plume]

Thank you so much, this is super helpful!! you are a star 🙌

[gray lily]

Hello everyone, last year I purchased a Rabbit R1. Today, I wanted to finally activate the Perplexity credit (see image), but when I click on the link, it says the URL is invalid. How should I proceed?

[regal turret]

[e-mail] — you should email here.

Looks like a bug.

[gray lily]

Will do, thanks

[charred fog]

Where is deepseek ? I can't rewrite with DEEPSEEK. Is it a bug?

[edgy spear]

Does anyone know the best way for Perplexity to output British English spellings and grammar?

Everytime I try to promt it to not use American English, it still goes ahead and spells everything wrong and I end up with needless corrections.

I'm mostly using claude model on Pro account.

[fluid lark]

It's given as "R1"

[charred fog]

I know "R1", but it did not display in "REWRITE"

[fluid lark]

Indeed. Probably a bug.

[pearl geyser]

did someone implemeted this awesome perplexity in trading? i would do also

[gloomy parrot]

chat, does anyone know if Comet will be free for Perplexity Pro users?

[gloomy parrot]

wait, fr, check the news -> chain of thought -> respond with actions | daily cron setup 👀

smth like that?

[pearl geyser]

i would like to implement some specified knowledge or reading the chart, using what i have learned, so to create a sort of application that send me specified alett on some spexified situation of market

[vivid vessel]

I subscribed to Perplexity Pro and selected Claude 3.7 Sonnet in the models section. Is Claude only getting used when using pro search (web/writing) or does it use Claude with auto (quick web/writing) search to?

[honest forge]

Ich habe mich jetzt lange mit Perplexity (Pro) "gestritten" 😆😠 und letztendlich nur widersprüchliche Informationen erhalten. Das in den Einstellungen ausgewählte Sprachmodell sei demnach wohl völlig irrelevant, weil jeder Modus, den man im Prompt auswählt (z. B. Auto, Pro, Deep Research, etc.) das eingestellte Modell überschreibt oder Perplexity im Hintergrund selbst entscheidet, welches Modell zum Einsatz kommt. Inwieweit ist das wirklich wahr? Wenn ja, warum wird so etwas nicht klar und transparent kommuniziert? Warum gibt es dann überhaupt die Auswahl der Modelle in den Einstellungen???

[vivid vessel]

Für Auto und Pro Search wird das ausgewählte Model verwendet

[vivid vessel]

Was noch wichtig zu erwähnen wäre ist, dass die Modelle nicht wissen welches Model sie sind, dass heisst, dass die Modelle sich als ein anderes (Halluzination) ausgeben können.

[honest forge]

Wie sicher ist die Aussage mit "Auto" und "Pro"? Ist das eine Information vom Support? Ich frage deshalb, weil ich durch mehrfaches sehr unterschiedliches Nachfragen und Recherchieren jetzt stets das Gegenteil mitgeteilt bekommen habe.

[vivid vessel]

Hab sie gefragt (Pro Support):
Hi there,
When you select Claude 3.7 Sonnet in your settings, it will be used for both Auto mode and Pro Search. The model you choose in settings acts as your default model across these search modes. The only exception is when you specifically select search modes that use their own models (like Deep Search with R1 or o3 mini) - in those cases, those specific models will override your Claude selection.
Let me know if you have any other questions!

[honest forge]

Danke!!! Bleibt die Frage, warum so etwas (aus meiner Sicht Wichtiges) offenbar nur einzelnen Usern bekannt ist. Schließlich führt eine solche Intransparenz zur falschen Handhabung von perplexity 🙃🫣.

[vivid vessel]

Ja, ich war mit dem Auto und Pro Modus auch verwirrt

[honest forge]

Aktuelle Antwort vom Support 🫣😉✌️

[vivid vessel]

Ich glaub mit Auto ist das Modell Auto in den Einstellungen gemeint

Also wenn ich Auto Modus verwende nutzt es immer Claude

Ich erkenne die Modelle am Schreibstill

Kannst es austesten in dem du Sonar eine Frage stellst und Claude

[honest forge]

✌️👋😉 hier die überarbeitete und ganz aktuelle Rückmeldung vom Support

[buoyant perch]

Does anyone know how to get rid of this banner? I clicked on enterprise pro just to see what it was, now I can't get rid of this banner cause apparently I started the sign up process, and it REALLY wants me to finish subscribing. I don't need enterprise. How do I get rid of this?

It's always on the top of my screen on every page while using Perplexity now.

[rustic thunder]

how can i contact perplexity to sponsor our hackathon?

[weak birch]

what kind of sponsorship? they don't give money for hackathons for advertisement

[deep forum]

Guys should i subscribe to grok through the IOS app or the website

[vivid vessel]

Perfekt und man kann auch auf dem PC sehen welches Model verwendet wurde in dem man das CPU Logo am Ende der Nachricht hält

[haughty brook]

#⚡│ask-community message

[whole veldt]

Anybody has an idea why spaces could stop working? Whatever space I try (or create new), the AI answers only the questions that i write and does not take into account the space instructions.

[vestal dock]

My team and I are building a Vercel webapp using the perplexity API. The app works in our development environment, but when we use it in production, our API calls are blocked by cloudflare for "bot protection". Does anyone know how to fix this?

Error fetching professor information:  Error: 401 <html>

<head><title>401 Authorization Required</title></head>

<body>

<center><h1>401 Authorization Required</h1></center>

<hr><center>openresty/1.25.3.1</center>

<script>(function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'919e31e034e587a1',t:'MTc0MDg5MDIwNC4wMDAwMDA='};var a=document.createElement('script');a.nonce='';a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();</script></body>

</html>


    at (node_modules/openai/error.mjs:44:0)
    at (node_modules/openai/core.mjs:292:23)
    at (node_modules/openai/core.mjs:336:0)
    at (src/app/api/stream-email-generation/route.ts:74:21)
    at (src/app/api/stream-email-generation/route.ts:384:32) {
  status: 401,
  headers: {
  connection: 'keep-alive',
  content-type: 'text/html',
  date: 'Sun, 02 Mar 2025 04:36:44 GMT',
  set-cookie: '__cf_bm=j3Nkrp771u2Vhvjo40.LazaudxxYik2rKT2ZI4hYxyI-1740890204-1.0.1.1-aZA4XHnTeE2uSWfGkascXci3flracFrFKrG09w3VatYHGLP7UnOsPjhA9DMNjd1XITnPOTJwEWxqHdgpa8URJobUpsbMPc_jzYR1_3eZw84; path=/; expires=Sun, 02-Mar-25 05:06:44 GMT; domain=.perplexity.ai; HttpOnly; Secure; SameSite=None',
  strict-transport-security: 'max-age=15552000; includeSubDomains; preload',
  transfer-encoding: 'chunked'
},
  request_id: undefined,
  error: undefined,
  code: undefined,
  param: undefined,
  type: undefined
}

[true whale]

Hi, how are you?
When I generate results in https://www.perplexity.ai/search, it returns good results.
But if I call same prompt via api, it does not correct results.
I used pro model in website and tried all models in api. But I couldn't get correct results.

How should I write api prompt to get same results?

Best regards

[ionic sparrow]

#🧪│api-general message

[spring anchor]

Issue on Win 10/11 for www.perplexity.ai:
I turned the default theme to dark theme in setting. After a few hours or a day, when I go on Perplexity, it will turn back to either system default or light theme. And I have to change the setting again. Why is that?

[rustic thunder]

api credits for our hackers would work

[vague stratus]

Is there a limit to the total # of sources it can get? Pro.

[fluid lark]

It only cites the first ten it sorts anyway

[sage echo]

hello! i came across the Perplexity AI Business Fellowship just now and am really interested in applying. i saw that applications ended 2 days ago, however, is there anyway to get in contact with the team to submit a late application?

[void jacinth]

Hi. Is it possible to disable the "People also ask" area that adds after each response of the AI?

[fluid ermine]

are there limits for deep research for pro?

[rich sage]

500 per day / You can view your quotas here: https://www.perplexity.ai/rest/user/settings

[ionic sparrow]

[e-mail]

[frail schooner]

why am i seeing ads in my paid perplexity account ? is that the correct behavior

or is there a way to turn it off

[lime crater]

does anyone know when the fellowship program starts? excited for it

[solar vine]

Where can i reach to any mod or someone to get a Pro role? I've texted a pro support arleady cause i havent received it when joining from the link in settings

[ionic sparrow]

https://discord.com/channels/1047197230748151888/1111786888626438245 hop in

[solar vine]

Thank you very much

[lofty vessel]

will there be an ai companion extension for firefox?

[elder escarp]

I’ve been charged twice who do I email

[ionic sparrow]

please contact [e-mail]

[kind tulip]

has anyone found a fix, for the ai when coding it sometimes messes up the layout

[frail schooner]

why am i seeing ads in my paid perplexity account ? is that the correct behavior
or is there a way to turn it off

[white hollow]

I also need my Pro role 😦

[grim pendant]

Hello everyone 🙂 Can you run Claude Code if you are on Perplexity Pro?

[rich sage]

No, this has nothing to do with Perplexity. For Claude Code, you need the Claude API (Anthropic Console). 😉
( https://docs.anthropic.com/en/docs/agents-and-tools/claude-code/overview )

[kind tulip]

u know how to fix htis ? its hella annoying

[formal forge]

Is this a place I can ask questions about perplexity?

[rich sage]

Do you mean that the code is not grouped into a single block?
If so, implicitly ask for it in your query:
"Respond in a single HTML code block" (or another format).

[kind tulip]

yes

ill try that

[still sonnet]

Is the file limit for spaces 500 or 50? The UI in perplexity says 500, support is saying 50.

[spiral falcon]

i would love to have feedback on which model i am using, a couple months ago i got pro and it says on the bottom right the model, have perplexity removed that?

[keen venture]

Hey guys did you notice that at the end of the message it doesnt show which model is being used? Used to show it there where the "i" is now

[cosmic rain]

Hi, any older post on how to generate PPT slides on Perplexity?

[whole veldt]

Can somebody add pro role for me? I left the server, rejoined from the pro link but I still can't access pro channels.

[spiral falcon]

it doesnt seem like perplexity web ui is in a great state now.... at least complexity extension fixes some features

if i deactivate the the web search action, does it matter if i deactivate pro search?

[spiral falcon]

i think it is back again

[grave berry]

Hey! I have changed my default assistant to Perplexity but whenever I hold the Home button I always get the Google assistant (Gemini) waking up instead of Perplexity. Any ideas? Im using S23U

[spiral falcon]

how can i get the pro badge? i joined through the web button but it doesnt work

[manic garden]

I set my default model to sonnet 3.7, and it appears when I open a new chat it is perplexity's default model. When I rewrite responses, I also don't appear to get a different model even though I select a new one. Any advice?

[pine hill]

a similar problem

[robust latch]

Hi All, I can not upload a document in Pro search, when I press the upload file button nothing happens whereas the upload photo feature is working fine. I have tried many times to sign in and out and closed app as well. But this feature isn’t working. Any idea how to fix this problem

[daring coral]

Hey all, been trying to initiate deep research using the API and I keep hitting 413 - Payload Too Large errors. Same prompt works in the chat interface on perplexity.ai and has worked just a couple of days ago when I initially tested my workflow.

Am I the problem or has Perplexity changed the limits / sizes for lower Tiers in the last few days? Any way I can test the size of my prompt? According to OpenAI's tokenizer the prompt has 2400 characters and >500 tokens.

Maybe worth mentioning that I am trying to set up an automation via n8n and their HTTPS node, so potentially also an error in how I configured the node?

[solid oyster]

Guys, is there any idea on how to speed up the reaction of the support in pro subscription?

@haughty brook

20 hours no reaction...

[ionic sparrow]

#👉│feedback-general message

[void jacinth]

There are some overall problems with switching models as I could find out using it about a week. Better to delete the response and make an input again cuz regenerate works strangely.

[warm swan]

Will the upcoming « Comet » got a professional tier compatible with the existing professional version of Perplexity ? Or will it merge with the existing Perplexity version ?

What do we know about it after all ?

[runic heart]

when using the "Pro" search, which AI model will be used?

[ionic sparrow]

#💬│general message

[spice ether]

Is it just me, or are the attached files in threads really annoying? They automatically get attached to the prompt every time I try to ask a follow-up question, and I have to manually remove them every single time. Why is that? I feel like that’s a very bad user experience design

[fluid lark]

i can only get the rewrite function to work on my phone

the mobile perplexity app

[cloud mantle]

I added 7 PDFs of 500 pages each in a private space

What response model do you recommend? I selected only my documents as a source, so it will only look for my answers in pdf files, right?

Is there anything you should know to best optimize this functionality?

[onyx oriole]

Is there a way to remove the news items from the Perplexity home screen?

[glossy pine]

Is there any chance I could post a google form link for a laptop survey? its for a school project and im trying to increase my sample size

[analog badge]

gemini i guess cause it has the largest context window, and you can mention only refer the document in the prompt

[molten quiver]

I'm not sure this is giving the full answer, or at least the current UI is making it extremely confusing to me. Just noticed recently that when using Pro search, the model is now 'Pro Search', which is not a model... I guess that it has some agentic behaviour (what search queries to launch...) but I was expecting it at least producing the last answer with your default model. I'm not sure though if this is the case or not. If you 'rewrite' with Sonnet 3.7, it still says the same.

[candid zephyr]

How can I use Claude 3.7 Sonnet in Perplexity Pro? I chose Claude 3.7 Sonnet AI Model in settings, but where can i find this AI Model to chat with it?

[fervent wedge]

Hi, Till recently I could use Perplexity.ai normally.
Now, I am getting this message:

Application error: a client-side exception has occurred while loading www.perplexity.ai (see the browser console for more information).
Any suggestions?

There is just that one line on a blank otherwise blank page.
I am on Firefox Browser on Windows.

I also noticed that the perplexity page opens, and right after that, within half a second it shows the above error on a blank page.

[brave flicker]

I created mulitple key for multiple client projects. but I don't see API key wise cost breakdown on the platform.
This make it difficult for me to bill my clients for their usage. Is there a solution ?

[ebon zealot]

Hello,

I've been trying to get in touch with the Perplexity team regarding startup credits but haven’t received a response yet. We’ve been using the Perplexity API for Better Buy and absolutely love it. While we have other offers for startup credits, I’m a big fan of Perplexity and would love to continue building with your platform.

Could someone guide me on the best way to connect with the right person to discuss this? I’m eager to make a decision soon and hope to keep Better Buy as part of the Perplexity ecosystem.

Thanks so much!

[hollow ginkgo]

Hey @ebon zealot!

If you find the original message helpful, please consider reacting to it with the :star: emoji. If the post is appreciated by the community and receives 5 stars, it will go to the https://discord.com/channels/1047197230748151888/1082806833938436228 channel and the post author will get the <@&1082034222778302614> role on Perplexity.

[lost zephyr]

1X

[haughty brook]

Please try again without using browser extensions it could be that one of your extensions is causing an issue on the website

[pine horizon]

Hi, I turned off the Pro Search button, but it keeps turning back on. How can I keep it off, I can't find it in the settings. macOS Perplexity app. Android app too.

[fervent wedge]

Hi, Thanks. The problem came up without any new addon being added, or even a modification to any addon's setting.
I did try turning off ublock origin and other other privacy ones (I have a few), and it didn't help.
I have way too many tabs open at the moment (and don't have a suitable way to get the windows in the right location on a restart - any suggestions?), so will turn off all addons on restart later.

[fervent wedge]

I wonder if anyone else is facing this issue.

[ionic sparrow]

reinstall browser

[fervent wedge]

I tried it in a portable version of Firefox I have. Same result 😦

[ionic sparrow]

did you try restarting your computer